Information identified as archived on the Web is for reference, research or recordkeeping purposes. It has not been altered or updated after the date of archiving. Web pages that are archived on the Web are not subject to the Government of Canada Web Standards. As per the Communications Policy of the Government of Canada, you can request alternate formats by contacting us.
Volume 5, 18 July 2011
TRANSCRIPTION OF PROCEEDINGS BEFORE THE CANADIAN RADIO-TELEVISION AND TELECOMMUNICATIONS COMMISSION
Review of billing practices for wholesale residential high speed access services. Notice of Consultation CRTC 2011-77, 2011-77-1 and 2011-77-2
140 Promenade du Portage
18 July 2011
In order to meet the requirements of the Official Languages Act, transcripts of proceedings before the Commission will be bilingual as to their covers, the listing of the CRTC members and staff attending the public hearings, and the Table of Contents.
However, the aforementioned publication is the recorded verbatim transcript and, as such, is taped and transcribed in either of the official languages, depending on the language spoken by the participant at the public hearing.
Canadian Radio-television and Telecommunications Commission
Review of billing practices for wholesale residential high speed access services. Notice of Consultation CRTC 2011-77, 2011-77-1 and 2011-77-2
Konrad von FinckensteinChairperson
Crystal HulleyLegal Counsel
James WilsonLegal Counsel
Tom VilmansenHearing Manager and Manager, Costing Methods and Tariffs
140 Promenade du Portage
18 July 2011
- iv -
TABLE OF CONTENTS
PAGE / PARA
1. CNOC-CORC (Canadian Network Operators Consortium Inc.)869 / 5267
2. Cogeco Cable Inc., Quebecor Media Inc. on behalf of its affiliate Videotron and Rogers Communications Partnership (collectively, the Cable Carriers)945 / 5706
12. Shaw Cablesystems G.P.1014 / 6145
4. British Columbia Broadband Association1037 / 6300
5. TELUS Communications Company1063 / 6450
6. Bell Aliant Regional Communications, Limited Partnership and Bell Canada1082 / 6582
- v -
PAGE / PARA
Undertaking998 / 6026
Undertaking1155 / 7040
- vi -
"Bill is now going t set the stage for the rest of our presentation."
"Bill is now going to set the stage for the rest of our presentation."
"The Commission has bee committed for"
"The Commission has been committed for"
"the facilities, ILECs, used"
"the facilities ILECs used"
"and though by many"
"and thought by many"
" hundreds and thousands of end users "
" hundreds and -- thousands of end users "
"They are Type II"
"There are Type II"
"megabyte per second"
"megabit per second"
Paragraph 2828, lines 13, 15-16:
Paragraph 2882, line 11:
--- Upon resuming on Monday, July 18, 2011 at 0906
5249 THE CHAIRPERSON: Good morning.
5250 This is the Rebuttal Phase. I think most of you are familiar now how it goes.
5251 We expect you to come back and give us your response in light of the submissions you heard, and then if somebody makes some contentious remark I may ask others to comment on it or if somebody feels that something was said by a particular panel is particularly egregious and needs correction, please put up your hand, okay, so we have a little bit of a dialogue as well as the presentations.
5252 Madame la Secrétaire, please start.
5253 LA SECRÉTAIRE : Merci, Monsieur le Président.
5254 Before we start I would like to go over a few housekeeping matters to ensure the proper conduct of the Rebuttal Phase.
5255 Please note that the Commission members may ask questions in either English or French. Simultaneous interpretation is available during the hearing. The English interpretation is on channel 1 and French on channel 2, and it can be selected from the left side of your microphone.
5256 For the public, you can get an interpretation receiver from the commissionaire at the entrance of the Conference Centre.
5257 I would like to remind participants that during the oral presentation they should provide for a reasonable delay for the interpretation, while respecting their allocated presentation time.
5258 We would ask that you please turn off, and not only put on vibration mode, all of your beepers and smartphones while in the hearing room as they cause interference in the internal communication system used by our translators.
5259 During this Phase, regardless of whether they are presenting oral rebuttal argument, participants may be called upon at any time to answer questions from members of the Panel in light of another participant's submission.
5260 For technical reasons and in order to facilitate the work of the Court Reporter and Interpreters I would ask participants that are not making presentations to identify yourselves every time you are asked to respond to questions from the commissioners.
5261 You are welcome to use the nameplate holders on your table to indicate your intention to speak to the Panel or simply raise your hand.
5262 It is also important that you turn off your microphone after each intervention in order to avoid interference in the audio system.
5263 Please note that there are also two microphones on stands for your convenience in case you don't have access to a microphone on your table.
5264 For the record I would just like to indicate that the order of appearance was amended on July 14th and revised on July 15th and was sent to all parties.
5265 Now, Mr. Chairman, we will proceed with the first presentation with CNOC, the Canadian Network Operators Consortium Inc. Mr. Tacit is appearing for CNOC.
5266 Please, Mr. Tacit, introduce your colleagues for the record, after which you have 12 minutes for your oral rebuttal argument.
5267 MR. TACIT: Thank you, Madam Secretary.
5268 Mr. Chair, Commissioners, we have the same panel that we had last time before you but the order at the table has been shifted a little bit.
5269 To my far right is Mr. Marc Gaudrault; beside him is Mr. Bill Sandiford; beside him is Mr. Stevens; and then to my left are Mr. Rocca and Mr. Andersen.
5270 Mr. Chair and Commissioners, in this rebuttal statement we will address three main issues:
5271 - the rate structure that should apply to wholesale high-speed services;
5272 - how the Commission can set fair rates for peak-demand usage; and
5273 - how to get from where we are today to the proposed rate structure without holding up competition.
5274 MR. SANDIFORD: After hearing the submissions of parties in the oral hearing to date, CNOC remains convinced that the 95th-percentile-peak-usage billing method is the one that most closely reflects how costs need to be recovered in incumbent networks. It is also the method that provides the greatest incentive for incumbents and independent ISPs to augment network capacity and increase their investment in order to avoid congestion.
5275 You have heard very clearly that all incumbents design and upgrade their networks according to peak demand and not traffic volumes passing through their networks.
5276 Once an independent ISP pays for all of the peak demand that it imposes on an incumbent network, the incumbent is fully compensated for the peak demand and for all levels of demand below that peak.
5277 Measuring peaks at the interfaces between incumbent and independent ISP networks is all that is required in both ILEC and cable carrier networks. At the end of the day, it is the total demand of an ISP that must be accommodated and paid for on a network, regardless of the ratios of downstream-to-upstream network capacity or demand. It is neither practical nor necessary for peak traffic demand to be measured at all the links within an incumbent's network.
5278 We also wish to stress that once an independent ISP pays for traffic on the basis of peak demand, it is compensating the incumbent for all the costs that peak demand imposes on all incumbent network links typically involved in handling that demand.
5279 Under our proposal it remains up to the incumbent to monitor and upgrade the links within its network as necessary to avoid congestion, just as it does today. Billing plans do not eliminate congestion, only proper network planning does. However, our measurement and billing method would generate sufficient funds that incumbents can use to augment network capacity required by competitors, if the incumbent is managing their network properly.
5280 Traffic patterns do not change unpredictably overnight. Measuring peak demand at the incumbent-ISP interfaces will give incumbents accurate, ongoing data regarding independent ISP peak demand. This will assist incumbents in their network planning processes and will reduce their risk in accommodating competitor capacity requirements.
5281 We are also prepared to reduce the incumbents' risk further by providing quarterly forecasts of peak demand. This is a reasonable co-carrier approach to risk management in a competitive environment and is consistent with the practice adopted in the local competition regime.
5282 There is another trend that is reducing the risk that incumbents bear in adapting their networks for competitor use. Total Internet traffic continues to grow and so the risk of capacity overbuilds is relatively low.
5283 Since independent ISPs are prepared to pay for peak demand, which includes all demand levels below the peak, gaming of the system is impossible.
5284 If peak demand usage is levied as a separate charge as we propose, the most appropriate rate structure for wholesale high-speed services is one that strips out all usage from the flat-rate component used to charge for access. The access charge should only recover non-traffic-sensitive access costs plus a 15 percent mark-up. It should not vary by speed unless and to the extent that it turns out that there are non-traffic-sensitive costs that vary by speed.
5285 The same principles should apply to the charge for the port fee at the ILEC or cable carrier interface.
5286 All usage-sensitive costs should be recovered through a 95th percentile peak usage rate. For DSL, this rate will essentially be a transport rate, since access costs are non-traffic sensitive. In the case of cable, the usage-based rate will recover the usage-sensitive -- and hence speed-sensitive -- access costs and transport costs and therefore a blended mark-up is required. The mark-up on the access costs would be 15 percent and the mark-up on transport costs would be a market-based rate.
5287 Independent ISPs of all sizes can adapt to this approach and deal with traffic spikes and transitional issues. The approach can and should be applied to all traffic and is easily auditable by independent ISPs, unlike a volume-based approach to charging for usage.
5288 Transitions from ILEC ATM to IP-based networks and from bridged to PPPoE arrangements can also be accommodated by independent ISPs of all sizes, provided that:
5289 (1) the costs of switching are kept modest;
5290 (2) competitors do not have to pay for two sets of interfaces at the same time during the transition; and
5291 (3) related service charges associated with facility changes are waived.
5292 Bell has already demonstrated that it is willing to adopt this approach since it has been actively seeking to move competitors from ATM to IP-based technology for several years. The need for other ILECs to do likewise is similar so they can also accommodate higher FTTN speeds.
5293 Some have criticized the peak-demand-usage-based charge approach as being harder to audit than a volume-based approach, but this could not be farther from the truth. It is straightforward for incumbents and independent ISPs to monitor traffic on their respective sides of their network interfaces. These interfaces number in the hundreds, whereas billing on the basis of volume requires incumbents to measure traffic for hundreds of thousands of end users. The ISP has no way to audit those volume measurements.
5294 This brings us to the second issue, which is how to actually come up with a fair rate for a peak-demand usage charge.
5295 Of course, we advocate a cost-plus approach based on Phase II principles. However, in order for the Commission to be able to come up with a rate, it must get a sense of what a reasonable mark-up should be on Phase II costs.
5296 In order to do that we have, in response to the second undertaking we gave last Tuesday, proposed a method of developing an average rate per incumbent territory that can be used as a reference point. That charge is based on the prices levied in the incumbent's territory to aggregate regional traffic and then transport it to a central point in the incumbent's operating territory. That rate relative to an incumbent's Phase II costs can be used to test whether a mark-up proposed by an incumbent over Phase II costs and the corresponding rate that flows from that calculation is reasonable.
5297 When it comes to a cable carrier, the rate may be different than that of the ILEC sharing the same territory, not because the mark-up would be different but because usage-sensitive access costs would also be included in the costs to be recovered through the peak-usage charge.
5298 We also recommend that this exercise be conducted annually so that incumbents' usage-based rates for wholesale high-speed services are reduced as technological advances and other factors reduce the underlying costs used to derive the rates.
5299 The third issue that needs to be addressed is how to get from where we are today to the new regime that we have proposed.
5300 Measurement according to the 95th percentile at the interface does not seem to be an obstacle for incumbents. As they acknowledged in the first round of this hearing, this is a method that they use in other parts of their businesses.
5301 Their main concern is that it may take a while for them to be able to develop billing software and make necessary back office adjustments to accommodate this form of charge.
5302 Rather than debating this issue, we propose an interim regime whereby incumbents will measure according to this method and simply bill this one line item manually per ISP until they modify their systems to handle the billing automatically.
5303 MR. TACIT: 95th-percentile-peak-demand capacity is the most efficient and competitively neutral manner of charging for usage.
5304 A volume-based approach constrains demand and actually blunts incumbent incentives to invest. All it does is protect the incumbents' TV businesses while disadvantaging those of competitors.
5305 As it stands, CNOC already has serious concerns regarding Bell's claims that its IPTV network is completely distinct from the network used to carry Internet traffic.
5306 Before concluding, I just want to note that while we did not have enough time to make specific comments to address Commissioner Morin's charts regarding the MTS Allstream model in our rebuttal statement, we are prepared to deal with that issue at this stage.
5307 We thank you for your attention and welcome your questions.
5308 THE CHAIRPERSON: Thank you very much.
5309 On paragraph 4, the last sentence, you say:
"It is neither practical nor necessary for peak traffic demand to be measured at all the links within an incumbent's network."
5310 That goes diametrically against what Bell said. Bell on the first day showed me their chart and said: Here, you're only measuring at the end but actually congestion occurs in all sorts of other points and you have to measure there in order to get a true reflection of the usage.
5311 How do I reconcile these two statements?
5312 MR. TACIT: So the exercise really amounts to knowing what all the costs are and then setting a rate, and in any exercise like that, unless you have a rate for each end user that's different, which is not really possible, there will have to be some form of mathematical approximation or algorithm applied.
5313 All we're saying is they can and should measure capacity at their links but that's for internal network-planning purposes, it's not for the purpose of billing per se.
5314 When it comes to billing, as long as we ensure that all of the costs are properly counted and that the peak rate is meant to recover those costs, the rest will be looked after by the statistical averaging of having so many ISPs affecting the network. It's not just the traffic of one ISP that's going to make the network fluctuate.
5315 I don't know if -- did you want to add anything to that statement, anybody else?
5316 THE CHAIRPERSON: Bell, it sounds logical what they're saying. Do you agree?
5317 MR. BIBIC: We completely disagree, Mr. Chairman. We'll cover this in more detail in our rebuttal statement, but here's the issue.
5318 Think back to last week. Vice-Chairman -- well, Mel Cohen from Distributel was on and he explained that if his peak is at 10:00 p.m., he admitted that it could very well be that in the aggregation network of the network provider that there's congestion -- I think he said 4:00 p.m. or 1:00 a.m. He gave two different points.
5319 The Vice-Chair asked him: Would you be contributing to the congestion at those other times even though it's not your peak? Mr. Cohen said: I very well could be and that's not reflected in my rate.
5320 The point being that when there is congestion in the aggregation network at times other than the ISP's peak, which they may be causing, we can't bill them for that unless we measure at each and every link, which isn't practical. But when our customers do it, we can. We do bill them for that extra usage.
5321 There was an exchange between yourself, Mr. Chairman, and Matt Stein of Primus where you kind of tried to cover this, and Primus' response was: Well, don't worry about what happens in the aggregation network at points 1, 2 and 3, that's up to the ILEC to manage.
5322 Yes, it is up to us to manage, but when our customers cause congestion, we will them; when their customers do, we can't, and 95th won't address it because it only measures a peak at one point in time, at one point in the network.
5323 MR. TACIT: So, as we're saying, the issue is recovery of total costs and application of peak on an industry basis. You cannot take a snapshot and say, well, at this one point in time this one ISP had an impact on this link over here that may not have been measured. It is an aggregate model. That is its strength and it has to be that way because it's not practical to use any other method.
5324 On the other hand, if you look at it from a volume perspective, it doesn't do absolutely anything for managing that capacity constraint either.
5325 MR. GAUDRAULT: In Mel's example, for example, at his peak, at 95th percentile he's paying for 95 percent of the peak, including everything underneath it. So if he has another peak at some other whatever, we're counting all of that. His peak is higher than whatever other micro-peak there may be somewhere down the line. He's paying for all of it.
5326 MR. SANDIFORD: To further that comment, it is our belief that if we believe Mr. Bibic's example that at some point there could be a small peak that they are not able to accurately bill is for, our point is that we are already compensating them at all the other times when our traffic is below our peak.
5327 So if you take a look at our graph -- and historically the graphs look like this and the peak is somewhere at the top -- there are a lot of areas underneath that 95th percentile line that our traffic has not hit. For example, we probably don't reach our peak 80 percent of the time, but we are paying them as if we reach that peak all of the time.
5328 So could there be a small peak at one point, at some point in a day that exceeds above our line and we cause them some traffic congestion in their network, I'm not sure, but I can tell you that we are already paying them for 100 percent of the time for the traffic at our peak, even though at many other times we are nowheres near that line.
5329 MR. TACIT: And there is one other final point.
5330 If you look at our response to Undertaking No. 1 and the interrogatory response that's attached, you will see that when an incumbent is dealing with more than one ISP it actually over recovers overall for peak capacity.
5331 Do you have anything to add on that?
5332 THE CHAIRPERSON: Mr. Bibic, you disagree?
5333 MR. BIBIC: Well, so if rates are truly based on costs -- and, by the way, you will hear when we come up that we are prepared to accept that our AVP rate should be based on cost. So if rates are truly based on costs and the costs are truly done properly and the assumptions are correct, we have heard throughout all of last week that AVP and 95th percentile should lead to the same revenues, should be revenue neutral.
5334 So if that's the case, I'm sitting here and I'm wondering why is CNOC so adamant that it be 95th percentile? Why?
5335 There is a reason. The reason is by paying 95th percentile, and only that, there is an incentive -- go back to Figure 2 in our opening statement last week, there is an incentive for if you are going to hit the same peak as somebody who has high usage, you might as well drive your usage up, too.
5336 So the point is that the ISPs with 95th percentile will be able to flatten up their traffic distribution to the highest point possible, not pay any more, but for us it will drive massive costs in the aggregation network. That's why they are so adamant.
5337 THE CHAIRPERSON: Why would it drive massive costs in the aggregation?
5338 MR. BIBIC: Because, Mr. Chairman, if you have a traffic distribution that's like this and you hit a certain peak, you might as well get it up to the highest level possible at all times of the day. That's going to cause congestion in other links that we hadn't provided for and we won't be able to recover that.
5339 Again, to just go back -- and I will cover this in my rebuttal -- if you go back to Matt Stein's testimony, he said: I'm going to make the most efficient use of what I pay for. And the most efficient use is to flatten that out. And he's going obviously try to flat -- or any small ISP --
5340 THE CHAIRPERSON: Sure.
5341 MR. BIBIC: -- is going to try to flatten it up to the maximum level possible. You are not paying any more since you have paid for the peak.
5342 But meanwhile there are links in the aggregation network -- points 1, 2, 3 in our chart Figure 1 -- that now become congested in a way that we hadn't provided for and that wouldn't be reflected in Commission cost studies.
5343 So if the rates are truly based on cost -- well, the rates will be based on some cost forecast that the Commission will allow. They will be revenue neutral, but it won't be -- 95th percentile won't be cost neutral to us. It will be revenue neutral, not cost neutral, because 95th percentile will drive more cost.
5344 AVP or the volume pricing, it's simple, it just tracks usage rates or usage billing to usage. It couldn't be any simpler.
5345 THE CHAIRPERSON: Mr. Stevens, I see you shaking your head.
5346 MR. STEVENS: I was shaking my head, Mr. Chairman, on a number of things.
5347 I mean the implication that we can dictate to our customers that we are going to flat line our traffic so it's equal all day long makes no sense. I mean every carrier in the room, I'm sure their peak in April was when the royal wedding took place, nothing to do that we as customers can say: You have to use the Internet at a certain time of day. It just doesn't happen.
5348 MR. TACIT: But again we come back to the basic point that we pay for everything under that peak demand. So it's a rate setting exercise and we shouldn't be having to manage their network.
5349 But at the same time the important points as to why -- you know, Mr. Bibic asked why is it important, there is three points.
5350 One, it doesn't constrain capacity, it allows users to use whatever they actually need to use in terms of Internet;
5351 two, it allows competitors to differentiate their services better without having to adopt a Bell-like volume-based charging model; and
5352 three, it's easily auditable.
5353 So those are the reasons.
5354 THE CHAIRPERSON: Now, when Mr. Stein from Primus was testifying, he had sort of a fallback model. He basically called it the pipe.
5355 Could you, CNOC, live with that?
5356 MR. TACIT: Mr. Rocca is going to address that.
5357 MR. ROCCA: So we still believe that 95th is a better model because it actually reflects the actual costs that are imposed in the incumbent network.
5358 The problem with the essentially flat rate pipe is that it overcompensates the incumbents for usage that is not put onto the network. It also imposes some additional issues for redundancy. For example, say there was an ISP that needed 500 megabits of aggregate usage, if they wanted to have two redundant gig links, they would actually be paying for 2000 megabits of capacity when they were actually only putting 500 megabits of capacity on the network.
5359 That said, any capacity-based model is better than any volume-based model, but we strongly believe that the 95th better reflects the costs that we are imposing on the network.
5360 THE CHAIRPERSON: In light of what Mr. Bibic just said about congestion maybe being caused at other points which would not be compensated, would this model not partially address this point?
5361 MR. TACIT: I don't really think it necessarily does any better than the other. In fact --
5362 THE CHAIRPERSON: Well, no. Mr. Rocca just said it overcompensates. If that's the case then --
5363 MR. TACIT: Well, sure you can overcompensate, but that's what -- we are trying to avoid that.
5364 THE CHAIRPERSON: No, no. But to that extent it's presumably benefit to Bell.
5365 MR. TACIT: Of course.
5366 THE CHAIRPERSON: You deal with those congestions.
5367 MR. TACIT: But the exercise isn't to overcompensate Bell, in our view, it's to compensate them fairly and fully but no more.
5368 THE CHAIRPERSON: Mr. Bibic...?
5369 MR. BIBIC: So in fact that model, Mr. Chairman, if you take the answers I have given you so far, the Primus model is worse. It's worse.
5370 So here's why. I think the CNOC point about overcompensating is a function of if you are a really small ISP and you need to step up to an extra increment of capacity, you have to buy the whole pipe and pay for the whole pipe even though you are not going to be using anywhere near the whole pipe. I think that's their point.
5371 But put the small ISPs aside and think about the answers I gave thus far, with peak pricing -- let's say your peak is at 80 -- I'm just using an artificial number -- at 80, but your pipe has a maximum of 100. With peak pricing the ISP has an incentive to drive its usage to 80 at all times of the day. With Primus' model, it now has an incentive to drive it up to the full capacity since it's paying for it at the end of the day. So it has an even greater incentive to flow volume through the network causing congestion in points 1, 2 and 3.
5372 And I want to make clear -- and there was an exchange the first day where people are interpreting what I said as somehow trying to squelch usage of the Internet.
5373 We want people to use the Internet, but we want -- and this comes back to an exchange Tom Little and I had with the Vice Chair -- we want people to use the Internet, we don't care if it's a grandmother with low usage or a very high user, as long as they pay for usage.
5374 THE CHAIRPERSON: Candice, over to you.
5375 COMMISSIONER MOLNAR: I have no questions, Mr. Chair.
5376 THE CHAIRPERSON: Len...?
5377 COMMISSIONER KATZ: I need to understand why you are not in favour of the MTS model, the Primus model. There has been very little discussion about it.
5378 I guess I come from the position that we, the Commission, have already recognized there is a need to create competition, more competition in order to protect Canadians, and facilities-based competition is not yet here. So it's our job to find a vehicle to create that competition and, in the simplest terms, it is to create an environment where broadband would be made available to a third party through a lease arrangement.
5379 So if I look at it at that level -- cost aside, and we will do the costs separately -- the obvious and simplest solution is to create an environment where you folks, and others who want to be in the business, have access to capacity that you need in order to serve your customers and after that the market will take care of itself. And it's the least intrusive model.
5380 And it goes back to the policy direction that was issued to us several years ago, and you folks are in the market and competing with the ILECs, the cablecos in a free and open market, subject only to the fact that we have created an environment in the simplest terms, in the most least intrusive way of giving you access to what today is not readily available financially to you, and yet you stay away from that and you say you would rather have more sophisticated models.
5381 I understand capacity costing, I think incurring the business risk that you would incur of having to decide how much capacity you need to serve your customers and then managing it is the best way of serving Canadians and giving them a competitive marketplace.
5382 MR. TACIT: Can I answer the question?
5383 Okay. So I have a couple of things to say about that.
5384 In terms of consistency with the policy direction, we actually think that's something that's already market based and exists out there is probably more closely reflective of the real world than this type of capacity pricing.
5385 Now, as I said, we do prefer a capacity pricing model over any kind of volume-based approach, but there is also the issue of efficiency and rate impact and let me give you a tangible example.
5386 MTS Allstream has always taken this approach to its A-HSSPI costs and with regards to legacy rates the net result was that for a 1 gig A-HSSPI people needed to pay $68,500 a month. Now, I can assure you that at that level the objective that you are pursuing, Commissioner Katz, of facilitating competition, just didn't happen.
5387 You know, if you look at the percentage of ISPs operating in MTS territory with that kind of monthly pricing, you will find that there is very little competitive activity.
5388 COMMISSIONER KATZ: I said we will deal with cost aside going forward.
5389 Everybody seems to be in favour of cost-based rates so that historical situation you are talking about, from my perspective, is not something that we need to rely on going forward.
5390 We will establish cost-based rates in all jurisdictions whether it's for DSL or TPIA going forward. The issue only is what model do we use that creates the competitive climate that this country wants to see, that we all want to see, to protect consumers, while at the same time letting the market take care of itself to the extent possible.
5391 And you don't have, unfortunately, a lease versus buy model for yourselves right now because you can't build your own network so you have to lease it. The question only is to what extent can you lease that capacity that you need in order to provide competitive services to Canadians.
5392 MR. TACIT: So I'm going to pass it over to my colleague in a minute, but before I do I also want to mention one other thing.
5393 The $68,500 example was an extreme example, but the point I was trying to make is that a monthly rate that requires ISPs to over pay for capacity -- and that's what we are really talking about. Let's assume we can even track the costing and bring it down, you know, accurately reflect what the costs are year-over-year and that the Commission will do a good job of monitoring those yearly and adjusting prices, let's assume even if that happened, there is still an overcapacity -- an overpayment for the same capacity issue.
5394 The fact of the matter is that it's simply not necessary to do that in terms of a -- I mean what you seem to be talking about, Commissioner Katz, if I'm understanding you correctly, it's really a risk allocation issue. I mean that seems to be your central concern.
5395 Do I have that right?
5396 COMMISSIONER KATZ: My central concern is being least intrusive -- using words from a policy direction -- while at the same time creating an environment for competition in this country to protect consumers.
5397 MR. TACIT: Well, the only intrusion would be having to adjust some billing systems and back office arrangements. The rest is done commonly in the industry so we actually think that's more consistent.
5398 MR. GAUDRAULT: I have an example here, if I could.
5399 So TekSavvy, we have 20 1 gig links and we have been telling Bell for a long time now we would love to go to 10 gig ports. So let's say we are using 18 gigs, we need to buy a third 10 gig port. So in this case we would be paying for a full 10 gig port that we are not using. So that's a big problem.
5400 COMMISSIONER KATZ: But they ILECs and that cablecos are building networks that they may not be using initially, but will need at some point in time.
5401 MR. GAUDRAULT: But this is the case, that I'm not causing 30 gig of traffic on their network, I'm causing 18. That hasn't happened. They have given me a port and as it gradually goes up I'm going to start using it.
5402 So the issue is that, you know, charge me for what I'm using. So the 95th is the equivalent to looking at the flowthrough, the pipe, and like if you could just see it, oh, there's the peak. Here's the whole pipe and they are using half of it, okay. Right? So that's the least intrusive. That's what 95th amounts to.
5403 Usually you top off the 5 percent just because there's little squiggles, you know.
5404 But that's effectively what 95th is, it's the least most obvious -- like to tell you how much are you using of that pipe. If you were to give me a rough estimate of that pipe, how much you are using, that is what 95th does.
5405 COMMISSIONER KATZ: I guess what you need to convince me of --
5406 THE CHAIRPERSON: MTS has a hand up.
5407 COMMISSIONER KATZ: I just want to make one comment.
5408 I think what you need to convince me of anyways is that this Commission and the industry won't get bogged down in reconciliation, in billing disputes, in audit disputes in perpetuity, because that's the fear that I have. I have seen it already before the Commission on a number of other situations as well where there are disputes and we have been forced to get involved in them and it's a no-win.
5409 And it's a very costly exercise and at the end of the day the person paying that cost is the consumer, because those are all costs borne by the industry that they have to pass back down to their consumers in order to deal with it. You get a bill, it's not exactly what you measured, you want to understand it all, there is someone getting paid to audit these things as well, and it just goes on and on.
5410 So the idea that I put forward, that MTS puts forward, is a simple one: you guys assume the business risk, you order what you want, we make sure it's cost-based -- I think everybody in this room has agreed to a cost-based model -- and then life goes on.
5411 MR. ROCCA: So the 95th is already -- you know, we were talking about less intrusive, the 95th is an industry standard, widely accepted model for shared network charging costing. I mean it's not something that's new that we have invented.
5412 The big problem with the MTS model is that when you order a gig of capacity, it's not like they go and augment all of their network saying: Okay, somebody bought a gig of capacity, now we need to go augment these links. They are still going to do their network capacity planning and only augment those links as they approach capacity. It has nothing to do with the amount that we have pre-ordered. It's not like they are going to go and pre-build capacity based on what it was that we order today.
5413 And you multiply that times redundant links and again, you know, I take you back to the example I used before where you are using -- you are paying for 2 gigabytes of capacity and using a quarter of it and you are paying for that whole 2 gigabytes, even though you are not imposing that and they haven't gone and added 2 gigs of capacity to their network to satisfy that.
5414 MR. TACIT: In terms of auditability it's actually the simplest of all because you can synchronize your clocks on both sides of the interface and measure exactly the same thing. It's actually the volume-based that's hard. Once this is established it's actually quite easy to implement on an ongoing basis because the industry does it widely.
5415 THE CHAIRPERSON: Mr. Rocca, that's twice now you have made this redundant point. I don't get it.
5416 If you were building rather than leasing you would take the business risk of building the redundancy or not? If it's not used, you have spent a lot of money for something that you don't need to.
5417 Why should it be any different when you are leasing it? I don't get this.
5418 MR. TACIT: Well, what we are striving to do is come up with the most economically efficient model. So we are saying that if you accept. first of all. the premise that the 95th percentile is the most efficient way of billing, then let's talk for a moment about risk.
5419 In terms of risk allocation, first of all, traffic patterns don't change overnight. Traffic overall is increasing. Incumbents serve ISPs in the aggregate so they are looking at an aggregate demand. The impact of anyone ISP coming or going or having good times or bad times is going to be limited in terms of any kind of risk impact. We have offered them forecasts, quarterly forecasts, which is the same type of approach, by the way.
5420 THE CHAIRPERSON: I asked Mr. Rocca about his question of redundancy and he basically says: If I have to use a redundant line I will have to pay for it even if I don't use it. I say that's the risk of business, you have that whether you build or lease. I don't know why because you are leasing you should get a better risk factor.
5421 MR. ROCCA: Well, actually, if I was to deploy my own fibre I would have multiple fibres in the same bundle. It would actually not cost me any additional income whatsoever to provide redundant capacity over that same link. I would be able to light up multiple fibres in that bundle.
5422 MR. GAUDRAULT: There is also the fact that upstream, if I could, upstream to us, so we are sort of in the middle if you will, downstream facing the customer we have the big ISPs, we have, you know, Bell and Rogers and that. Facing the Internet we have what we call transit providers that we pay. So this is the exact method we use with them. TekSavvy has two of them, we all have them. Every single ISP period, including Bell, including Rogers, they all use this method to do business.
5423 MR. SANDIFORD: Mr. Chair, with regards to your direct question about the redundancy on the links, we are saying that if we need redundancy on the link because we have customers and we need to have the high level of up time, we are happy to pay for the facility for the second link, meaning the actual physical link charge itself.
5424 What we are talking about here today is billing for usage. I shouldn't have to buy 200 megabits of usage because I want to serve 50 megabits of traffic to the customer.
5425 If there is a cost that is to be borne by the incumbent by building me the second physical link, I'm happy to pay them for the second physical link. Why shouldn't I, I have asked for it.
5426 But in terms of billing for the usage, under the MTS Allstream model, which the question was, if you charge me and you forced me to pre-buy usage at the same rate as the link, now I'm buying usage in addition to the link in orders of magnitude greater than I need.
5427 COMMISSIONER KATZ: You are not buying usage, you are buying bandwidth.
5428 MR. SANDIFORD: Under the MTS Allstream model all of the usage that would be accounted for or that we would use is built into the price of the cost of that link. So as opposed to billing on a volume-based model like Bell's AVP model, they are suggesting that the usage is built into the cost of the link.
5429 THE CHAIRPERSON: Let's hear from -- we have MTS, we have PIAC and then somebody in the corner there. Okay, Mr. Cohen.
5430 But first, MTS.
5431 MS GRIFFIN-MUIR: Thank you. Teresa Muir.
5432 I just want to clarify -- and I think Mr. Katz actually made a good clarification -- you are buying bandwidth at the port, so even though someone is saying, "Can you open the port so I can have, let's say, a gig, or in the example of TekSavvy 10 gig", the ability of you to pass through greater traffic through that port starts the day you get the port. So we, MTS, have to provision for that eventuality.
5433 So in essence you are asking us to provision and then just measure your usage. That's why we developed the model the way we did and that's why we are saying you are sharing the risk.
5434 THE CHAIRPERSON: PIAC...?
5435 MR. LAWFORD: Mr. Chairman, we did not have our hand up.
5436 THE CHAIRPERSON: Sorry, behind you. Behind you sorry, not PIAC.
5437 MR. LAWFORD: Thank you.
5438 THE CHAIRPERSON: It was the gentleman behind you.
5439 Go ahead.
5440 MR. BHULLAR: Hi. This is Jatinder Bhullar.
5441 I think the issue is that in traditional data you have a port charge and a user charge and from what I understand from CNOC is they are willing to pay the port charge if they want to build redundancy based on costing, but that redundancy doesn't increase the usage by itself so the network of providers don't actually -- when a link fails and somebody uses a redundancy, it doesn't have any impact on their network, just they have to -- as normal, if they didn't buy redundancy they will still be provisioning to volume or traffic.
5442 So redundant links need to be treated separately like an interface charge or an interface access charge, but the usage is the true load, which doesn't change whether you use the redundant link or, you know, the TekSavvy link fails so they move the traffic to the other link.
5443 So it is very important. This traditional data has been done for ages, that you buy a port and then you buy usage or capacity.
5444 Those components are easily manageable and configurable and they have virtually no network impact beyond the point of that interface.
5445 Thank you.
5446 THE CHAIRPERSON: Mr. Cohen...?
5447 MR. COHEN: It's Mel Cohen, from Distributel.
5448 First, to Commissioner Katz's concern about the billing disputes, I just want to point out that Distributel has purchased Internet transit from MTS Allstream, from TELUS, from Bell, from Videotron and from Shaw. In every case they have billed us on the 95th percentile and we have not been embroiled in billing disputes.
5449 Every one of those carriers is capable of 95th percentile assessment and, as CNOC had mentioned, the application in our case for where we want to measure usage of the network is maybe 200 different ISPs that have to be measured. So this is not a massive job to implement 95th percentile.
5450 But, to the question of the Allstream model, what I want to point out is best by example. I'm going to give you the example of Distributel's interconnection with Rogers for TPIA in Mississauga.
5451 We have about 700 megabits per second of traffic. About six months ago, seven months ago we ordered a second GigE link into that same POI anticipating future growth. It did actually take six months to get it installed, by the way.
5452 So we have got this GigE now. It's in there. The traffic is still just at around 700 megabits per second. We do pay for that second GigE. We made a choice that we wanted additional capacity and we are paying for a link that runs from Mississauga to Front Street at the same rate we paid for the first one. That's the additional capacity that we purchased and we are paying for and we are happy to pay for that.
5453 But then you take the point where it connects in Mississauga and you look at the traffic in the cable network, the distribution network that we were causing back and forth. That hasn't changed.
5454 So the problem with this model is that it links the physical size of the connection and tries to use that to recover the usage on the network and that's just not right.
5455 And there are some cases where you -- you know in our case it was anticipating growth. There are other cases where an ISP will want to have a second provider for diversity and redundancy. You could want to have three different providers connected and spread 200 megabits worth of traffic over 3 gigabits worth of backhaul capacity. You just don't want to have to pay for that usage on the Cable Carriers' network based on the size of the interconnection point.
5456 THE CHAIRPERSON: Okay.
5457 Len, back to you. Len, did you have any other questions?
5458 COMMISSIONER KATZ: No, that's all for now, Mr. Chairman.
5459 THE CHAIRPERSON: Tim, did you want to add on?
5460 COMMISSIONER DENTON: Mr. Cohen, forgive me this. This may be completely obvious to you. Can you just run over -- you said the problem with the Allstream model is it links the physical size of a connection to the recovery of costs of usage, or words to that effect.
5461 Can you just go over again why that is bad in your view?
5462 MR. COHEN: Because it requires us to make a decision that will impact such issues as redundancy and recoverability to usage of our customers of their facilities. The fact that I want to have three different paths back from Mississauga to Front Street doesn't impact the capacity that they have to have available to them in their network. I'm not using that.
5463 I mean, remember we are separating -- we are trying to recover the costs of their -- you know, their build from their POP out to their customers. That's what we want to compensate them for. And the size of the port that we interconnect with doesn't reflect that. It's -- what we want to do is we want to put the 95th percentile measurement at that point which actually will reflect how much we are using.
5464 Does that answer your question?
5465 COMMISSIONER DENTON: In other words, the size of the link at that particular point is an extremely imprecise or useless proxy for what the real costs that need to be recovered. Is that correct?
5466 MR. COHEN: Yes. I would say it is very imprecise and there are reasons why you wouldn't want it to be anywhere near the same.
5467 COMMISSIONER DENTON: So, in other words, as a method of recovering the costs causally associated with usage, the size of the interconnection link is more than imperfect. It is not desirable.
5468 MR. COHEN: Yes, it's a very poor proxy.
5469 COMMISSIONER DENTON: Okay, thank you.
5470 MR. TACIT: Could I just add one quick comment --
5471 THE CHAIRPERSON: Hang on. Just hang on, please. Only one person can direct traffic here. TELUS has -- let's hear from them and then we will come back to you.
5472 MR. MANSOURATI: Thank you, Mr. Chairman. I would like to make a comment about opportunity costs.
5473 When we provide an interface to a customer, in this case an ISP, we have to mention the network in such a way that we should be prepared to satisfy the demand when it comes on that interface. Therefore, there is an opportunity cost that comes up in this argument and we have to be prepared at any given time to provide the connection in a well and reliable operating way.
5474 So it's important to consider the opportunity costs. That's the point that I am making.
5475 THE CHAIRPERSON: Back to you. Did you want to say something, Mr. Tacit?
5476 MR. TACIT: All I wanted was to say was just adding to the point that Mr. Cohen made, that having a port size doesn't tell you anything about what the capacity is at any given point in time that's being pushed through that.
5477 If an ISP gets a second link for purely redundancy reasons, the overall capacity impact on the network will be the same as long as all the traffic is going through link A until it's severed and then it has to be shifted to link B. It doesn't change. The total capacity is the same.
5478 And in terms of sizing networks, carrier size network is based on the overall demand that they experience over a period of time. This doesn't get adjusted second by second, minute by minute depending on somebody ordering a GigE today. It is a long term planning process and traffic patterns are well established especially in the aggregate.
5479 THE CHAIRPERSON: Tom, you had a question?
5480 COMMISSIONER PENTEFOUNTAS: Yes.
5481 Is there a possibility of -- going back to something that Mr. Tacit said and touching on something Mr. Bibic mentioned earlier, could we miscalculate peak? If peaks vary at different points in the network could we be mistaken in calculating peak usage if we are simply measuring it at the interface?
5482 MR. TACIT: I don't think it's likely in the aggregate that that could happen. As we said, could there be a mismatch between a particular node and a particular interface at a particular point in time for one particular ISP? Sure, because you know we are talking about trying to de-average something, a model that inherently is constructed based on averages.
5483 The model works because it's an aggregated model and if you try and change the fundamental mathematical underpinnings, of course it may not work that well, but that's not what happens in the real world.
5484 COMMISSIONER PENTEFOUNTAS: But could that throw a wrench in network capacity planning, if you excuse my perhaps improper --
5485 MR. TACIT: Well, I think the best example of that is that transit in transport providers can do it. They have multiple customers. They don't just have one ISP customer. They have multiple customers. It's not different.
5486 What is different is that there are multiple links. But in terms of the thought process and the nature of the exercise, it is the same type of exercise that's being done and it works today in the real world.
5487 COMMISSIONER PENTEFOUNTAS: Someone want to add something to that?
5488 MR. GAUDRAULT: In the aggregate basically if you are telling me that if I buy 5 or 10 or 20 gig links that that equates to everything I'm signed up for, they are telling me that that's how they manage their network by the port speed that I have.
5489 But the reality is that if you look at the access network, the port speed coming into the network from the customer's end, if you were to add up all those DSL lines together it adds up to way more than what the transport network is able to do. They are building their own network based on the likelihood of the usage going through it.
5490 COMMISSIONER PENTEFOUNTAS: One question: Have you calculated what the cost of developing and building software and the back office management would be?
5491 MR. TACIT: We can't do that for them. They have to do it for themselves.
5492 COMMISSIONER PENTEFOUNTAS: And who would pay for those costs?
5493 MR. TACIT: I would say to the extent that it's an incremental causal cost, the industry would have to bear it in some manner. It would have to be done in a reasonable manner over time so as not to cripple the very competition you want to foster.
5494 But you know we accept that if those costs are real and adhere to Phase II costs, the industry would have to pay them.
5495 COMMISSIONER PENTEFOUNTAS: Thank you, sir.
5496 THE CHAIRPERSON: Okay. And we have got two intervenors, Mr. Cohen(ph) first and then the gentleman in the corner. I'm sorry. I can't read that far away.
5497 THE SECRETARY: Please turn your mike on and say your name.
5498 MR. MOORE: Yes, sorry. Andrew Moore.
5499 I just want to point out on the issue of per-link billing that that could be easily solved on a per-link commit by the independent ISPs. So they would basically prepay for the expected usage that they think they are going to pass -- that that's going to pass on that link.
5500 And there would be no penalties for guessing wrong. They would just basically pay a wee bit more beforehand. Instead of at the end of the billing cycle they would pay at the beginning and if it's not being used it could be recouped in the next month.
5501 So that's just a suggestion.
5502 THE CHAIRPERSON: Marc, you have some questions?
5503 COMMISSIONER PATRONE: Just a couple. Thank you, Mr. Chairman.
5504 Just wanted a note of clarity, Mr. Bibic. You were saying that as far as capacity-based models go, you believe that the 95th percentile model is the lesser of two evils in comparison with the MTS/Primus model?
5505 MR. BIBIC: In terms of the costs that it will cause in the aggregation network through increased usage that we would not recover, yes.
5506 Again, it's back to imagine you buy a pipe that at maximum throughput can, you know, handle 100 units but your peak is 800 units. If you are paying for peak pricing you have an incentive to drive up your capacity to 80 units as many times of the day as possible. If you are going to pay for the full pipe regardless of what you use, you might as well try to drive traffic up to 100 since you are paying for the full capacity at all times of the day.
5507 Flattening your traffic distribution to its highest level possible with capacity pricing or costing, the highest capacity possible is the full size of that pipe that you are paying for rather than the peak that you are paying for.
5508 COMMISSIONER PATRONE: But the idea of driving up usage, I mean, doesn't that depend on your customer base? If they don't have the customer base that would warrant that type of usage then how do you simply drive up the usage?
5509 MR. BIBIC: By the way, in terms of the responses to the MTS model I do agree with much of what CNOC has said and Mr. Cohen has said.
5510 So if you don't have that customer base then as an ISP you are paying for far more than you are using. But the larger ISPs will have an incentive to adopt any kind of marketing plans or tactics that they can possibly implement in the marketplace to drive up and flatten their traffic distribution.
5511 Imagine like a midnight madness sale, you know, midnight madness. Any downloading or any activity on the Internet between midnight and six a.m. totally free regardless of what you do. That might displace some people and cause them to undertake some activities at those times of the day that they otherwise wouldn't, which is going to cause congestion on multiple links in the aggregation network for us, costs for us to augment those links that won't be recovered.
5512 So whether it's 95th percentile or capacity pricing à la Primus, the incentives are the same, just one you know has a more dramatic effect on us than the other but they both have the same impact.
5513 COMMISSIONER PATRONE: And yet, if I understand Mr. Tacit, what you are saying doesn't entirely make sense.
5514 And I'm trying to reconcile the two polar opposites, you know, that are looking at this model and saying it won't work for basically different reasons because he is saying that it's going to overcompensate the network operator. So if you are being overcompensated then wouldn't that be a bad thing from their vantage point and a good thing from your vantage point?
5515 MR. BIBIC: If you are a small ISP that's forced to buy far more bandwidth than you need, yes.
5516 If you are a larger ISP that can handle -- you know, that has the customer base already, and there are some large ones here sitting in front of you that have the traffic volumes and since they are going to have to pay for capacity anyway now -- I mean Mr. Stein basically said it last week.
5517 I want to use the bandwidth that I buy the most efficiently as possible. Mr. Stein wouldn't come forward and be a benevolent ISP and say to Bell, "I'm willing to pay you for the total capacity of the pipe just because I'm a good ISP and I want to be friendly do you".
5518 No, he is going to buy that capacity and if he is going to be paying for capacity regardless of what he uses. A good businessman has an incentive to make sure he uses that pipe as efficiently as possible.
5519 COMMISSIONER PATRONE: Okay.
5520 MR. BIBIC: That will drive costs to us that are not reflected in the cost model.
5521 COMMISSIONER PATRONE: So you are saying you won't --
5522 THE CHAIRPERSON: That's where you lose all of us. They paid for it. Why can't they use it 100 percent? I just don't get this.
5523 MR. BIBIC: Mr. Chairman, look, the bottom line is this: If you fully cost that pipe and you force the ISPs to pay for the full cost and you do the costs properly with no magic markers in the costing group, the wholesale price will be higher than retail.
5524 And as I sit here, in my heart of hearts, and you all know it; you will never let the wholesale price be higher than the retail price -- ever. We will not recover those costs. And if those costs are higher than the retail price you won't let it happen.
5525 All the ISPs came forward last week. CNOC said it, Primus said it, Mr. Cohen said it. They said they want cost-based rates but, but make sure that those wholesale prices are below retail and you give us a sufficient margin to let us compete.
5526 So if you could sit here and guarantee us that all those costs will be recovered that would be one thing. But it won't happen.
5527 Our rate, our wholesale AVP rate if you translate that --
5528 THE CHAIRPERSON: Okay, stop. That is a totally different point until now. You are now saying the costing rules don't work, which is a different story.
5529 We were talking here about the principle of them buying capacity from you. If they buy for a certain capacity why can't they use it fully which is what the Primus model says? And you say that's -- now, you have switched horses. You are basically telling me our costing rules don't work.
5530 MR. BIBIC: No, Mr. Chairman, I haven't switched horses. What I said throughout this whole morning is that 95th for capacity pricing will cause costs in the aggregation network which we won't recover.
5531 I have said it six times today. It will cause costs in the aggregation network to augment links that we otherwise hadn't provisioned for because of the incentives on the ISPs to drive up usage. Yes, to accommodate the fact that they are paying for the pipe they will drive up usage, costs will be created in the aggregation network. We can't recover those costs.
5532 That's the fundamental problem. This is where 95th and Primus model completely breakdown.
5533 MR. TACIT: So can I make a couple of quick remarks?
5534 THE CHAIRPERSON: Hang on. We have got an intervenor in the back there.
5535 Identify yourself, please, so people --
5536 MR. BHULLAR: Mr. Chairman, this is Jatinder Bhullar.
5537 I believe I think we are having issues here in terms of costing that Bell is saying they are not going to recover in the aggregation network.
5538 We have got to remember that what goes through to the ISP is the only traffic that is carried. Anything that happens either as congestion or in the aggregated network or whatever, that traffic doesn't get served. It becomes a traffic engineering issue to ensure that whatever they have committed to the ISP at that interface gets delivered. And some places it will be requiring augmentation.
5539 Plus, I think the issue is they are moving all their IPTV traffic to a separate network and claim a separate network. So they should be happy as anything that they are not going to have forward, you know, congestion in their aggregation network because those customers are moving to a separate network.
5540 So they should be like happy as anything to incur more ISPs rather than make everybody like look like a Bell Internet customer. Hey, if you do not behave like a Bell Internet customer, well, you know like we have to have really fees to ensure that we enforce really a punitive discipline on the ISPs.
5541 It just doesn't make any sense to me anyway. So thank you.
5542 COMMISSIONER PATRONE: No, I had my --
5543 THE CHAIRPERSON: Okay.
5544 Tom, you have a question?
5545 COMMISSIONER PENTEFOUNTAS: Yeah. I just have to get back to Mr. Bibic.
5546 Because to the Chairman's point and that's what we were discussing earlier, the point that I don't understand is, the ISPs are paying for that capacity. Once we have established they are paying for that capacity the worst case scenario for you, they use that capacity. They have paid for it. The best case scenario for you they come in under that capacity and you come out ahead. Your network has some leeway there.
5547 And we discussed last week the fact that it's pretty hard to get -- you know to have a steady stream at your capacity throughout the month or the year.
5548 So again, anywhere we are dense -- maybe I'm dense but where is the problem? They are paying for it. They should use it and they probably won't be able to use 100 percent of the capacity they paid for at any rate is my understanding.
5549 Tell me where I'm wrong.
5550 MR. BIBIC: Okay, Mr. Pentefountas, here is the problem.
5551 We are having a theoretical discussion and we operate in the real world. So this is what's going to -- if you translate our costs, our AVP, the cost, but if you translate our AVP costs into a megabit per second price which is what 95th percentile or the Primus model would require, you are up to about $39 per megabit per second.
5552 CNOC has filed an answer this morning, saying that they are only prepared to pay between 50 cents and five dollars. If you slice -- if the Commission slices our costs in half, we are down to about $20 to $22 per megabit per second.
5553 At those wholesale rates, and you add access to that, you are going to have likely an overall wholesale rate that's higher than retail or that gives the ISPs a very small margin. The Commission will not let that happen.
5554 COMMISSIONER PENTEFOUNTAS: It's a pricing issue.
5555 MR. BIBIC: Well, we won't recover our costs.
5556 COMMISSIONER PENTEFOUNTAS: Back to the Chairman's point about the horses or the horse, it's a pricing issue. It's not the fact that they are going to buy this capacity and their model of normal business practices if you bought it -- if you paid for it you should use it. Otherwise, you are coming -- you are losing. I mean that's not your issue.
5557 The issue is the price -- it comes back to the actual price. And aren't we in a situation where we are going to calculate costs with a mark-up and at that point if they are buying back capacity and they are paying for it -- like the issue is not the capacity. The issue is the actual price at the end of the day.
5558 MR. BIBIC: And it will be established by the Commission at an artificial level that ensures that there is a margin. There has never, ever, ever been a situation certainly as long as I have been doing this, where a wholesale rate was allowed to be higher than a retail price.
5559 Now, of course, it's been allowed for Videotron and MTS in some cases but it will not -- and supposedly even those are very rare. It won't happen. It just won't happen.
5560 MR. DANIELS: If I may add here, if you look at it, and I do not want to get into the details of the costing model, but at a high level the costing model that we do assumes that we are using a shared network. It's looking at costing that shared network because we are not selling end GAS neither to individual ISP end-users nor to our retail customers' dedicated use all the way through the network. And no on, no one is suggesting that.
5561 However, what is being suggested specifically with the MTS model here is that you -- that we cost as if that happens, that we cost, that you bill -- because I buy the whole capacity I can use at 24/7 and that's how you, Bell, should set the per cost model to get approved by the CRTC and that will be the actual rate that comes out of it.
5562 And that's why we are saying if you take that, the costing in terms of the cost model, we filed our megabit per second rate for 95th percentile and it's the same cost. Just multiply it by -- as if you use the full capacity the full time. The costing is the $39 done.
5563 So if you look at that and you multiply that out, that's how come we are saying it won't work in terms of it won't serve your purposes because they will be paying for capacity as if it was throughout our whole network which is not the underlying assumption. It's a shared network.
5564 THE CHAIRPERSON: Okay. And Mr. -- sorry.
5565 COMMISSIONER KATZ: I just want to comment on this issue.
5566 Mr. Bibic, unless you can prove to me that the quality of your customers are different and the quality of any of your wholesale customers, when you alleged that wholesale rates would be higher than retail rates, I come to the conclusion that you are then selling below cost because your costs are high as well and you are allocating those costs proportionately, presumably, whether it's your customer or whether it's a CNOC customer or a Distributel customer.
5567 What am I missing?
5568 MR. BIBIC: The point here specifically, Commissioner Katz, is that we are talking about buying with the suggestion of the MTS model is that you are buying for the full capacity as if you are using it 24/7, not on just the link connecting the ISP to our network, that is if you are costing it as you are buying it throughout the entire network on a shared basis. So our proposals and the way we have approached our costing is to treat it as a shared network so that it's the exact same service that we would provide our retail customers, it is the same. Our costing is done the same.
5569 But the suggestion is, and why we are saying it's going to drive up our costs, that the usage that's going to be driven is going to drive further usage on our network in the way that the cost model doesn't pick it up.
5570 COMMISSIONER KATZ: But your notion of it's the exact same customer being given the exact same service is where one of your flaws is, because what we are trying to do here is create competition and create distinctions and differences between the providers out there, not just more "me too" customers.
5571 MR. DANIELS: And absent --
5572 COMMISSIONER KATZ: What we are saying is there needs to be competition in this country and if it can't done through facilities-based, then I guess we need some regulatory intervention. And if we go that way the issue is, how do you create that competition while still letting the marketplace, to the extent possible, serve the public?
5573 MR. BIBIC: Mr. Vice Chair, I'm going to rise above from the costing detail.
5574 Another problem with the MTS model that hasn't come forward yet is that it makes a mockery of speed matching.
5575 We went through an arduous hearing last year where the commission decided that mandated access to higher speeds was necessary and it would be through speed matching. The MTS model says: Here you go, you get a 32 megabit per second pipe, do what you will with it. So now the ISP can offer any kind of speed up to 32 megabits per second that wants, regardless of what we actually offer in the marketplace.
5576 Anyhow, I didn't think that the scope of this hearing was to revisit speed matching and I would say that it would be a perverse outcome that we, the ILEC who invests, are not allowed to launch a speed at retail unless we make it available at wholesale, but now the wholesale ISP could take our network and offer any speed it wants. It's the same issue that we had with ADSL-CO but at another place in the network.
5577 So, you know, if we are going to have that debate now where the ISP can differentiate itself to that point, well, that wasn't what this hearing was supposed to be about and whether its 95th percentile or AVP -- and we do not support 95th percentile -- but if it's either one of those and you are paying for what you are using, then, you know, the ISP can fashion whatever pricing plans it wants.
5578 It was the very first question that the Chairman asked me on Monday last week: Let's be clear, with AVP can an ISP fashion any pricing plan they want? And the answer was yes.
5579 THE CHAIRMAN: Mr. Sandiford, you had an interjection.
5580 MR. SANDIFORD: Yes. I want to comment on some of the comments that my colleagues at Bell made to Mr. Pentefountas' question.
5581 I heard a couple of things that jarred my mind. I heard the comments "artificial", I heard Mr. Daniels twice say "it's our shared network".
5582 Telecom Order 2007-20 set the rates for Bell Canada's Ethernet Transport Service, Ethernet Transport being effectively what Bell uses to transport the various different backhaul throughout their network.
5583 We heard Mr. Bibic say that if he took their AVP rates -- and I also heard the word "artificial" somewhere in there -- come out, that it translated to a rate of somewhere in the order of magnitude of $39 per megabit, whereas CNOC was being unreasonable in our assertions that we had suggested a rate of $.50 to $5.00.
5584 In Bell's Ethernet Transport Service Tariff, which came out as a result of 2007-20, as a Category 2 competitor service -- so not even an essential service, it has Bell's appropriate mark-ups -- the rates for transport of a shared network were effectively set at $.35 on a shared basis.
5585 If we went to the assumption that we were going to use a dedicated network and we said, no, Bell, we want you to carve out -- we don't want the shared network, we want you to dedicate bandwidth to us period, the rate on that basis was only $10 per megabit per second, yet we have already heard Mr. Daniels say -- not once but twice in the last five minutes -- that it was a shared network.
5586 We don't understand where this $39 and the AVP rates come from and how that has cost-based and we believe it's clearly artificial.
5587 THE CHAIRPERSON: Well, at the end of the day we are going to hear from Bell, so let's hold our horses on that part.
5588 Rogers, I believe you had an interjection.
5589 MR. ENGELHART: Thank you, Mr. Chairman.
5590 THE CHAIRPERSON: You have to identify yourself because I'm getting e-mails that people listening to us online don't know who is speaking up. So both people in the audience and the Commissioners should identify themselves.
5591 MR. ENGELHART: Thank you, Mr. Chairman. Ken Engelhart from Rogers.
5592 We are up next so we were figuring we could set all this straight when we got up, but it seems that the discussion has gone completely off the rails so I thought I would just try and jump in at this point.
5593 I think the reason for the confusion or the reason for the way people are talking at cross purposes is CNOC is looking at the distributed access networks as though they are a pipe. That's the imagery that we are using, a pipe. They are saying the way we buy pipes, like transit pipes, is we pay for the 95th percentile. That's the way transit pipes are bought and sold.
5594 They are absolutely right. The 95th percentile is a good way of describing a pipe. If the distributed access network was a pipe they would be right.
5595 The reason we are having this debate is because the distributed access network is a series of pipes and that's where the confusion is coming in and that's where we are talking at cross purposes. So to use a slightly imprecise analogy, they are saying, look, we have pay for the traffic on Main Street so now we can put as much traffic as we want on Main Street and Elm Street and Maple Street. Well, wait a minute, no. You have paid for Main Street, but Elm Street and Maple Street are going to cost us money to augment and you are not paying for that.
5596 That's what this whole debate is about, is Bell is trying to make the point that in a distributed access network it's not a pipe and if you try and describe it as a pipe you are making a mistake. So when they say we are paying for the whole capacity why can't we use the whole capacity, that's just not correct when you are not talking about a single pipe.
5597 That's where this discussion has gone off the rails. That's I think where Bell is saying: Look, if we end up charging you for Main Street and Elm Street and Maple Street you are going to pay a lot more money. And I think Mr. Katz is saying: Well, why would the two customers be any different? They are really not right now.
5598 Our concern is, if you move to this pricing where you are only paying on Main Street, they are going to incent their customers to use Maple Street and Elm Street more and our costs are going to go up.
5599 THE CHAIRPERSON: Okay. Thank you very much. That's a very graphical way of explaining that. I appreciate that clarification.
5600 I think, Michel, you are the next one.
5601 I see all sorts of flags. My Commissioner has been waiting very patiently to ask some questions on the MTS model. Let him ask and then I will pick you up, okay.
5603 COMMISSIONER MORIN: Thanks, Mr. Chair.
5604 In paragraph 2 you wrote that your model is the one which provides the greatest incentive for incumbents and independent ISPs to augment network capacity and increase their investment in order to avoid congestion.
5605 Why, from your point of view, does the MTS proposal not provide those incentives?
5606 MR. TACIT: Because it's not really responsive to the actual amount of capacity used. It actually over buys. We have no idea how the incumbent is going to use those funds. They may augment, they may dividend out, they may do something totally different with it. There is a break in the linkage between the cost recovery and the funding that's required to actually augment capacity because we are overpaying.
5607 COMMISSIONER MORIN: But if they are overpaying they will pay for the capacity they want and they will take the chances that the capacity won't be there. So they have to forecast their demand and then increase their capacity with some caution, you know.
5608 So in that sense the MTS model, it seems to me, it will spark more investment in order to make sure. If the small ISPs want to offer more options to their own customers, they have to have the right capacity.
5609 MR. TACIT: So you heard the TekSavvy example where they need 18 gig and they would be forced to buy 30 under this model and pay for that. It's a cross subsidy. MTS Allstream could use that to lower its prices on its IPTV or its retail offering.
5610 We have no guarantee that if they over recover for capacity that money is going to be invested. We do know that if you create anti-competitive subsidies they will be used anti-competitively at the end of the day. Competition law and economics tells us that.
5611 So I don't buy the argument that forcing someone to over buy upfront something that they can't use is going to necessarily translate into a bigger investment in the industry.
5612 The other thing is, even if they were to invest, from a purely economic perspective they wouldn't necessarily be investing in the right places to meet consumer demand. The economic signals that this sends are not proper.
5613 COMMISSIONER MORIN: Thanks.
5614 Those are my questions, Mr. Chair.
5615 THE CHAIRPERSON: Okay.
5616 A couple of flags that came out before. Okay, let's start with the far corner.
5617 Please identify yourself.
5618 MR. MOORE: Andrew Moore.
5619 We have heard that nice analogy about Main Street and all those small streets, but in the case of an independent ISP what is happening is that all traffic at the end ends up on Main Street and has to exit by Main Street, which is why 95th percentile is actually accurate at measuring the impact of an independent ISP on somebody's network.
5620 THE CHAIRPERSON: Can you repeat that? I'm sorry.
5621 MR. MOORE: Okay. Basically we have heard the example of Main Street and traffic going through all those little, small streets around it, but in the case of an independent ISP the traffic all ends up on Main Street at the end. It has to go through the ISP's POI, which is why 95th percentile as measured at the POI is actually accurate at measuring the impact of that independent ISP on the incumbent's network.
5622 THE CHAIRPERSON: Well, if I understood the analogy that Mr. Engelhart offered, it is the ISP may have -- the traffic that they buy, they may be bringing it from Main Street onto Elm Street and thereby causing congestion on Elm Street because the ISP's customer lives on Elm Street and so therefore that's not being picked up in the 95th percentile.
5623 MR. MOORE: But at the end of the day the traffic all ends up at the same place and calculated. If there is a peak or the 95th percentile happens to be at a certain rate per second, a certain rate of bits per second, that's what the ISPs are paying. Regardless of where the congestion is located they are still paying for it. They are paying for what they are using.
5624 THE CHAIRPERSON: Yes, but the point is that they are causing congestion on Main Street and Main Street needs to be upgraded and that's not being picked up.
5625 That was Mr. Engelhart's point.
5626 MR. MOORE: It still passes by that street. If there is congestion at that street and there is ISP -- there is traffic from that ISP passing on that link, it's still going to end up passing at a POI where it's going to be calculated and accounted for.
5627 THE CHAIRPERSON: Yes, but I thought Bell and Mr. Engelhart both made the same point, you are not picking up the costs that you are causing in this graphic illustration to the upgrading of Elm Street, you are only paying for the upgrading of Main Street. Elm Street you are not. It's not caught in this model and that's a flaw in the 95th percentile.
5628 Mr. Engelhart, did I misunderstand you, or Mr. Bibic, but that's what I understood was essentially the point both of you were making.
5629 MR. ENGELHART: Exactly right. And the intervenor is correct that -- you know, I could have drawn a more elaborate analogy where all of the roads funnel back to a single point, but the point is that when you are dealing with a network of roads they can have different traffic jams at different times and at different points and simply looking at the point where the exit to the freeway doesn't capture the state of congestion or the state of traffic on those different roads and that's why a 95th percentile measure is a good description of a single pipe and a bad description of a network of pipes.
5630 THE CHAIRPERSON: Yes. Okay.
5631 I will come back to you, because I see another flag way back in the corner there.
5632 Yes, identify yourself.
5633 MR. CHRIS ALLEN: Hi there. Chris Allen with the B.C. Broadband Association.
5634 I think the analogy is a useful way of looking at this and here is another way of looking at it using streets.
5635 I would say that independent ISPs are much like a small town off the side of a freeway, the freeway being the actual pipe that the ILEC delivers. In the 95th percentile the idea would be to stick the toll booth for the highway on the on ramp onto the highway and look at rush hour and say: Okay, at rush hour there are this many cars going through per second, let's charge the ISP based on that for the entire 30-day period of the month.
5636 What I understand is the ILEC's argument with that is that the actual traffic jam occurs farther down the highway, maybe three or four cities down the highway after many other on-ramps have entered the on-ramp and they have to build that capacity to increase the size of the highway. But the ISP feels that it's paying its fair share by paying not only for every car that goes across, but paying for the rush-hour peak that is going across that highway and compensating fairly by taking care of its own customers, paying it's fair share of the further down highway upgrades.
5637 THE CHAIRPERSON: Okay.
5639 MR. BHULLAR: Mr. Chairman, this is Jatinder. I have a couple of points.
5640 I think Mr. Engelhart, when he states that there is a confusion or we are off the rails, I think there is no substance to that.
5641 We all understand that the networks are designed, including all links, and then they end up in, you know, one link on the customer side and an aggregated link at the ISP or to Internet or transit through the Bells and everything. So that's how networks are designed.
5642 Costing models, as Mr. Katz has mentioned, will determine what is a good cost for usage when the traffic leaves for the ISP and it will take into account anything between the ADSL or the Rogers modem and, you know, like their nodes to determine what is an effective usage state.
5643 So I don't think there is any confusion. Like that's how networks are designed. By just saying streets makes no sense.
5644 Point 2 is actually the Primus model, MTS Allstream model and 95 percent model, they can all be converged into a solution. The solution is that you can have a 1 gigabyte pipe for example, then if you rate limit it, say at 500 megs, it may become the MTS Allstream model, and if that 500 megs translates into a 95 percent usage level, then it becomes the 95 percent usage model solution, as well as what, you know, everybody is saying, the actual usage.
5645 So there is a way within technologies to design that interface in such a way that it will accommodate a convergence of 95 percent and rate limited solutions to match up with actual usage and everything will line up.
5646 Thank you.
5647 THE CHAIRPERSON: Okay.
5648 Mr. Leong...?
5649 MR. LEONG: Yes, hi. This is Munly Leong from the World Broadband Foundation.
5650 I would just like to point out that the whole off peak would drive them to augment costs for usage between 12:00 a.m. to 6:00 a.m. business is really just fear mongering.
5651 This is minority power use of behaviour, it's not the average person behaviour who would have to go out of their way to change their sleeping patterns, the usage habits.
5652 It's actually volume pricing which segments peak and off-peak that drives people to do this to gain every gigabyte they are able to get. If things were unlimited or the cap was large enough on a regular basis people would just go about their lives and use what they need to when they need to and they would prefer to do it at reasonable hours.
5653 THE CHAIRPERSON: Okay.
5654 Mr. Sandiford...?
5655 MR. SANDIFORD: With respect to your question, Mr. Chair, towards Mr. Engelhart about the Elm Street, Maple Street, Main Street model and how the peak capacity billing models adequately compensate for the cost of Main Street but not the side streets, under a peak capacity billing model let's just say that a particular ISP peak is 500 megabits per second, that typical ISP would probably only reach that rate for two hours of the day. So we are paying for 500 megabits per second on Main Street for two hours of the day.
5656 It's our assertion that the other 22 hours of the day when we are not driving 500 megabits per second down Main Street we are still paying for it. We are still paying for it all the time and those costs that we are paying for for 24 hours a day, but yet not using, could be allocated towards upgrading Elm Street and Maple Street.
5657 THE CHAIRPERSON: Okay. Thank you very much.
5658 Tim, you have a question?
5659 MR. TACIT: Mr. Chairman --
5660 COMMISSIONER DENTON: I have been waiting 20 minutes. I will pull rank, Chris.
5661 COMMISSIONER DENTON: In the earlier discussion there was a discussion on which Mr. Cohen and CNOC commented basically that linking physical size of connection to recover cost of usage was a bad idea.
5662 Now, my question is for Mr. Stein of Primus. Would you like to comment on the views advanced by Mr. Cohen and CNOC that port size was no proxy for usage?
5663 Would you care to comment?
5664 MR. STEIN: I certainly was planning to comment later, but certainly I can do that now.
5665 I do believe it is a reasonable proxy for usage, but it's certainly not as efficient as 95th percentile for any of the parties involved.
5666 If you recall, during our opening comments I said Primus is strongly in favour of a 95th percentile model, however, any peak capacity billing model is better than any volume model.
5667 So while this is perhaps a reasonable proxy in the short term, it's certainly not preferable. The ideal solution is the 95th percentile model. That's really it.
5668 COMMISSIONER DENTON: Thank you. Perfect.
5669 THE CHAIRPERSON: Marc...?
5670 COMMISSIONER PATRONE: Thank you, Mr. Chairman.
5671 Mr. Sandiford, you gave us an explanation about a minute ago about how you reached the peak about two hours per day, for the rest of the time you do not reach that peak but you pay for it, right.
5672 But earlier Mr. Bibic mentioned the midnight madness argument in which your members would, under certain circumstances, say under the MTS/Primus model, strive to increase volumes to that high level by offering cut-rate deals in the middle of the night, that kind of thing.
5673 Should this be something we should be concerned about?
5674 MR. SANDIFORD: So I can't remember the name of the gentleman from the World Broadband from a moment ago, but I think he addressed this, whether advertently or not, in saying that the number of customers that are willing to stay up for a midnight madness day are probably a minority case. What drives our true traffic is the majority of our customers.
5675 The traffic graphs in my actual ISP that I run have been the same for 15 years. They have grown, but the peak has always occurred at the same time of the day, it is a two-hour window.
5676 And I want to clarify that two-hour window. For some ISPs it may be one, for some ISPs it may be four hours, but the point that we are getting at is it's a drastically smaller portion of the day.
5677 So, you know, I would love to be able to and, you know, some ISPs have tried in the past to try to shift usage patterns to off-peak period, but we simply have been technically and not practically able to do it on the level that is being imagined by those who would suggest we are trying to game the system.
5678 COMMISSIONER PATRONE: But if you had some particular promotions that you could offer, conceivably you could use that type of offer to try and gain market share from, say, the network providers, say Bell.
5679 Is that not correct?
5680 MR. SANDIFORD: Well, with regards to gaining market share, that's what we are trying to do. With regards to driving costs on the network, I don't think that we could ever practically run some sort of midnight madness sale that would ever drive our off-peak usage anywhere near our peak usage.
5681 THE CHAIRPERSON: It's 10:30, we have been an hour and a half, this can go on for the whole day,. I think we have heard basically every point of view and every angle on it, unless somebody really feels there is something that is vital I'm going to call a break now.
5682 MR. TACIT: Just 20 seconds, if I could.
5683 THE CHAIRPERSON: Okay. Twenty seconds. You, then Bell, then Primus and that's it.
5684 MR. TACIT: Okay.
5685 So on the cost issue, those ethernet rates that Bell mentioned are also through that entire aggregation network and pick up Elm Street and all the other streets. So that's the first point.
5686 The second point is that we have to keep remembering that this model is an aggregated statistical model. If you try to use it for individual examples you can always find an example that won't work, but that's not the point.
5687 My third very quick point is Mr. Bibic mentioned the unlimited rate of 32 megabits per second for access.
5688 This isn't new. MTS Allstream has had this rate structure in place of not having tiers but rather offering the highest speed for access from the start. So we view that as being completely on the table for all incumbents.
5689 THE CHAIRPERSON: Bell...?
5690 MR. CONDON: Thank you, Mr. Chairman.
5691 I just want to say that I hope we will be looking forward in building the pricing model instead of backwards at past models.
5692 What is clear is that more and more services are being built that take advantage of cloud computing, such as backups, music photos, movies seem very likely that will drive traffic 24 hours a day and I think that that's what we are missing in this argument.
5693 THE CHAIRPERSON: Okay.
5694 Primus, last word.
5695 MR. STEIN: Very quickly, two quick points.
5696 First, Commissioner Patrone, to your question about gaming the system. If an ISP were able to devise some sort of a midnight madness model, such as they do in other jurisdictions around the world, well, isn't that an example of the innovation that we are talking about. When we look for these points of innovation, they are not always going to sparkle like the sun. Sometimes some of them are relatively simple but fit for particular set of users, such as the one you are describing. I think it would be great if ISPs could do it.
5697 The second model, this morning we are having a lot of conversations about capacity planning and not a lot of conversations about a charging model. We can't mistake capacity planning that will continue to happen in all networks from a charging model.
5698 To use the example of cars -- not exactly a perfect analogy, but the point was counting all those cars during peak is an effective way to count all the cars, regardless if they are on Main, Elm, Maple. Make a left on Elm, make a right on Maple, whatever it is, we are counting all the cars and that's the important point.
5699 Thank you.
5700 THE CHAIRPERSON: Okay. Thanks very much.
5701 We will take a 10-minute break.
--- Recessed at 1038
--- Resumed at 1053
5702 THE SECRETARY: Order, please. A l'ordre, s'il vous plaît.
5703 THE SECRETARY: Mr. Chairman, just for the record I would like to indicate that I have filed -- the two responses to the two undertakings given by CNOC were put on the public record.
5704 Now, we are ready to proceed with Cogeco Cable Inc., Quebecor Media Inc. on behalf of its affiliate Videotron and Rogers Communications Partnership (collectively, the Cable Carriers). Appearing for the Cable Carriers is Mr. Dennis Béland.
5705 Mr. Béland, please reintroduce all of your panel for the record, after which you have 20 minutes to make your oral rebuttal argument.
5706 MR. BÉLAND: Good morning, Mr. Chairman, Commissioners, Commission staff.
5707 My name is Dennis Béland and I am with Quebecor Media. I have with me today Pierre Roy Porretta from Videotron.
5708 We also have with us:
5709 - from Rogers: Ken Engelhart, Dermot O'Carroll, David White, David Watt and Ddin Morbi;
5710 - from Cogeco Cable: Michel Messier, Philippe Jetté and Mike Coltart.
5711 And we also have with us Suzanne Blackwell from Giganomics Consulting.
5712 Il semble y avoir unanimité entre les parties à cette instance quant au fait que non seulement la concurrence mais également l'utilisation de l'Internet sont gagnants lorsque les tarifs de gros reflètent fidèlement les coûts de réseau sous-jacents.
5713 Les principales questions qui restent à résoudre sont les suivantes:
5714 (1) Quelle est la façon optimale de facturer la composante utilisation?
5715 (2) Où doit-on tracer la ligne entre la composante accès de la composante utilisation?
5716 Une ramification importante de la deuxième question est de savoir si les coûts d'accès doivent varier en fonction de la vitesse d'accès, et si cela devrait être reflété par les tarifs d'accès.
5717 Nous aborderons dans le cadre de la présente réfutation verbale chacun de ces points.
5718 Nous formulerons également certaines observations additionnelles visant spécifiquement le modèle de MTS Allstream.
5719 Pour terminer, nous aborderons brièvement une série d'importants points secondaires qui ont émergé au cours de l'audience de la semaine dernière.
5721 MR. ENGELHART: We begin with the most contentious of the outstanding issues, namely whether usage should be billed on volume or on peak traffic.
5722 We agree in principle with Mr. Tacit's statement that "what matters in terms of network usage is the costs that relate to expansion of capacity." The debate is not with this statement but with determining the best proxy for recovering the augmentation costs that are driven by additional usage.
5723 The Cable Carriers' usage-based costs primarily involve augmenting capacity at neighbourhood nodes. The augmentation is done when the peak usage at the node reaches a threshold limit. There are thousands of nodes and they all have different peaks at different times of day. Moreover, these peaks constantly shift from one day to the next. This makes it completely impractical to bill for usage based on the peak traffic at the individual node. Perfection is not on the table. We need a proxy.
5724 The CNOC model would choose as a proxy the aggregate peak traffic at the point of interconnection, or POI, between the cable carrier and its wholesale customer. This is an idea that lands far from the mark in terms of approximating the true network cost of additional usage.
5725 Worse still, if adopted, it would create incentives that could result in substantial uncompensated increases in the very costs it is attempting to recover. As Commissioner Denton noted in a question "How you billed for it would have different consequences on the incentives of the third-party wholesaler."
5726 The aggregate peak traffic of a wholesale customer measured at the POI does not correlate well with the peak traffic in the individual neighbourhood nodes. As evidence, we attach as Schedule 1 a previously confidential interrogatory response. We release this information because this matter has become so contentious in this proceeding.
5727 The information demonstrates that for Rogers' network, for example, the aggregate network peak does not correlate well at all with the peaks at the individual nodes. The aggregate network peak is very highly concentrated in the late evening. By comparison, the vast majority of our individual nodes achieve peak utilization outside of that time period. Peak traffic at the POI is a poor proxy.
5728 As Schedule 1 shows, the Rogers network peak is coincident with the traffic peaks in only 13 percent of our network nodes. In other words, a tariff that charges only for network peaks correlates with only 13 percent of the cost drivers in our network.
5729 The Cable Carriers' proposal is to use traffic volume throughout the distributed network as a proxy for usage-based costs. This approach is vastly superior to CNOC's proposed measure.
5730 The total volume of traffic within any given node is very closely correlated to the peak traffic of that node. As evidence, we attach as Schedule 2 another previously confidential interrogatory response, which shows this stark correlation for Rogers in graphical form.
5731 Simply put, the total volume of traffic passing through an individual node tracks very closely the peak traffic in that node, which in turn drives the segmentation of the node. Traffic volume is an excellent proxy for usage-based network expansion costs. To use the road analogy, roads with more traffic tend to have more traffic jams at rush hour.
5732 Schedule 2 shows that with a correlation of 79 percent, the higher the volume of gigabyte traffic in a node, the higher will be the peak utilization. Therefore our proxy is based on a 79-percent correlation, whereas their proxy is based on a 13-percent correlation.
5733 The wholesalers argue against volume-based usage charges on the basis that they should not have to pay for gigabytes that otherwise "ride for free." Schedule 2 lays this argument to rest. On a highly reliable statistical basis, the greater the traffic volume passing through a given node, the greater the peak tends to be and the closer we find ourselves to augmentation.
5734 End users are using the Internet more and more. As an access provider, it is our job to do the augmentation which allows for ever-increasing traffic. However, wholesalers should pay for the traffic they generate and which drives usage-based augmentation costs. This will not happen if wholesalers pay only for peak traffic measured at the POI.
5735 Another important fallacy in the wholesalers' arguments against volume-based pricing is the suggestion that they are "only using what they already paid for." Under this school of thought, it would be entirely acceptable for a wholesaler to interconnect at a given aggregate traffic speed, say 100 Mbps, then provide incentives to its end users to maintain this aggregate traffic speed 24 hours a day, seven days a week across the entire network.
5736 Aggregate gigabyte volumes could easily double or even triple relative to historical volume distributions, imposing substantial incremental costs on the underlying network builders, but as long as aggregate peak traffic does not increase, no additional compensation would be paid.
5737 CNOC and others argue that measuring peak traffic at the 95th-percentile model is widely used for transit. They are correct. But this does not make it a good model for measuring usage on distributed access networks.
5738 The peak traffic at the 95th percentile is a good description of the size of a single pipe. It is not a good description of a distributed network with many thousands of pipes, each with their own peaks. It is misleading to suggest, as CNOC does, that by paying for peak traffic measured at the POI they are somehow paying for the full pipe to each end user they serve. There is no dedicated pipe from the POI to each end user; there are only shared pipes.
5739 The billing model based on peak traffic aggregated at the POI is conceptually incorrect for all the reasons given and that is our main reason for opposing it.
5740 However, there would also be significant implementation costs with introducing it. CNOC and others argue that billing for 95th percentile will be easy because it is so widely used. The fact that it is used for transit networks does not make it easy for access networks.
5741 It is not the 95th-percentile calculation that is hard but the application of it to our network and existing billing system. It would require us to make significant changes to our usage-based billing systems and have that integrated with our provisioning system and with our Operational Support Systems that would collect the raw data.
5742 The costs of billing system changes would increase the TPIA rate. The billing system would be only for the TPIA customers and not our own retail customers.
5743 Under Phase II costing principles, the costs of the billing system should be recovered solely from TPIA customers. All this is for a billing model that does not reflect the true underlying cost of the network and which would replace volume-based billing measurements that already exist, that are well understood and that are transparent.
5745 MR. BÉLAND: The split between access and usage is about allocating the non-usage-sensitive and usage-sensitive costs and recovering them according to their cost drivers. Done properly, this will send the right economic signals to network owners and wholesalers. Ultimately, this will result in fairness for end users.
5746 We think the access rate should be a flat rate per end user, as do most parties. However, we disagree that a single access rate should be applied to all cable wholesale access services regardless of speed. A single access rate would be entirely inappropriate for cable networks.
5747 Providing higher-speed tiers costs more because these services require a larger allocation of the shared capacity on our hybrid fibre-coax networks. Higher-speed services use more 6 MHz channels, more CMTS and more fibre. They also require more stringent plant conditioning.
5748 The other critical difference we wish to emphasize regarding the access and usage split is that there are more usage-sensitive costs in the cable network. We need to make sure that the usage charge captures the usage-sensitive costs. Setting an artificially low usage charge will drive up usage but leave us with no means to recover the costs of that higher usage.
5749 We are very concerned about comparisons of our usage rates to deliver traffic with the rates provided for traffic on point-to-point transit networks. None of these are apples-to-apples comparisons.
5750 Rates for transit services are different than for last mile because the services are entirely different. Transit services require building a very small number of very-high-capacity pipes, typically fewer than 10. In contrast, TPIA services require building a very large number of small-capacity pipes to every single home. Clearly the per unit costs of transit are much, much lower and comparing the two does not make sense.
5751 Even comparisons between companies' last-mile networks are not like-to-like comparisons. Each company has designed their network differently and, as a result, the costs are not the same. Forcing uniformity on the rates for usage among these networks would be wrong. It would mean some companies would never recover their costs.
5752 Il a beaucoup été question dans le cadre de la présente audience du modèle de MTS basé sur la tarification de la capacité maximale d'utilisation. Le conseiller Morin a notamment présenté un tableau dans lequel on retrouvait une comparaison entre le prix final de gros par utilisateur qui résulterait de l'application du modèle MTS et ceux qui résulteraient de l'application de divers autres modèles, incluant celui des câblodistributeurs.
5753 Le conseiller Morin a également souligné ce qui semble être les effets bénéfiques du modèle de MTS, y compris les mesures incitatives encourageant l'adoption par les entreprises de gros de comportements responsables en ce qui concerne l'approvisionnement d'interconnexion.
5754 Nous réfutons comme suit le modèle de MTS.
5755 Premièrement, pour ce qui est de la comparaison des prix finaux de gros par utilisateur qui découlent de l'application de chacun des modèles, nous constatons que ces prix dépendent fortement des hypothèses de base, en particulier en ce qui a trait au trafic moyen d'utilisation par utilisateur final.
5756 Le tableau du conseiller Morin suppose une utilisation moyenne de 100 gigaoctets par utilisateur final par mois. Toutefois, le Consortium, dans le cadre de la présente instance, a publiquement affirmé que les utilisateurs de gros finaux consomment en moyenne 30 gigaoctets par mois.
5757 Utiliser le chiffre mis de l'avant par le Consortium permet de considérablement réduire les écarts de prix qui se dégagent du tableau du conseiller Morin. C'est ce que les câblodistributeurs ont fait à l'annexe 3 de la présente réfutation.
5758 Comme vous pouvez le voir, le prix final de Cogeco, quant à lui, est maintenant inférieur à celui de MTS Allstream.
5759 Mais ce n'est pas tout. Chaque réseau d'accès a sa propre structure de coûts. Même à l'intérieur d'une catégorie donnée de réseaux d'accès, par exemple ceux des compagnies de téléphone ou ceux des câblodistributeurs, il est parfaitement raisonnable de s'attendre à ce qu'il y ait des différences significatives dans les coûts.
5760 Ces différences peuvent s'expliquer par les différentes topologies de réseaux, le choix de différents fournisseurs d'équipements, différents déclencheurs pour les activités d'augmentation du réseau, l'existence de grands projets de modernisation de réseau ou des conditions du marché local qui affectent les hypothèses de la demande.
5761 Aucun des opérateurs n'a accès aux études de coûts détaillées des autres opérateurs, et il nous est donc impossible d'établir des comparaisons directes concernant les hypothèses de coût. Seul le personnel du Conseil jouit d'un tel degré d'accès.
5762 Deuxièmement, nous saluons les efforts du conseiller Morin à attirer l'attention sur l'absolue nécessité d'instaurer des mesures incitatives encourageant l'adoption par les entreprises de gros de comportements responsables. Cependant, nous soumettons respectueusement que l'accent n'est pas mis au bon endroit.
5763 L'idée que les entreprises de gros devraient être responsables de la prévision et de la commande, en temps opportun, de leurs installations d'interconnexion n'a rien de nouveau ou de bien surprenant. C'est de cette façon que les choses se font de nos jours, et les câblodistributeurs travaillent en étroite collaboration avec les entreprises de gros afin de s'assurer que des installations adéquates soient en place, en coopération avec des tiers fournisseurs d'installations, aux endroits où ce type d'installation est utilisé.
5764 On retrouve à l'Ordonnance de Télécom 2011-256 un exemple parfait de ce genre de collaboration dans laquelle le Conseil a approuvé un processus consensuel flexible instauré entre Vidéotron et les entreprises de gros avec qui elle fait affaire quant à l'installation et l'augmentation d'installations Ethernet 10 gigabits.
5765 Là où nous cessons d'être en accord avec le portrait brossé par le conseiller Morin est lorsqu'il présente son hypothèse selon laquelle l'approvisionnement responsable des installations d'interconnexion équivaut à une utilisation responsable de nos réseaux répartis.
5766 Comme les câblodistributeurs l'ont souligné à maintes reprises tout au long de la présente instance, c'est tout simplement faux. La récupération des coûts basée sur l'utilisation uniquement via les redevances fondées sur la capacité d'utilisation maximale mesurée au point d'interconnexion crée une puissante incitation à accroître à l'intérieur de ce seuil l'utilisation agrégée. Cela pourrait en bout de ligne nous imposer des surcoûts non compensés substantiels lors de l'augmentation de nos réseaux.
5767 On ne peut pas dire que le système de facturation aura rendu les entreprises de gros plus responsables si ce système incite ces mêmes entreprises à ne pas faire un usage responsable de nos réseaux.
5768 Le Conseil devrait garder à l'esprit le fait que MTS est une ESLT au Manitoba, mais un grossiste dans la plupart du pays. Ils ont sans doute l'espoir que le tarif qu'ils proposent sera adopté non seulement pour les clients résidentiels, mais aussi pour les clients d'affaires. S'il est adopté pour les clients d'affaires partout au Canada, ils espèrent pouvoir bénéficier de l'application du tarif.
5769 En effet, ils peuvent très bien payer une contribution en tant que vendeur à Winnipeg afin de recevoir des subventions en tant qu'acheteur à Toronto, Montréal, Vancouver, Edmonton et Calgary, qui sont tous des marchés beaucoup plus grands.
5771 M. MESSIER : Le Consortium des Opérateurs de Réseaux Canadiens et Distributel se sont montrés préoccupés par l'impact de la décision du Conseil dans la Politique réglementaire de Télécom 2010-632, selon laquelle le Conseil a ordonné aux câblodistributeurs d'établir des points d'interconnexion agrégés.
5772 La décision d'exiger des points d'interconnexion agrégés, plutôt que des services basés sur la tête de ligne, était en partie motivée par l'établissement d'une symétrie entre les obligations réglementaires des ESLT et celles des câblodistributeurs. Le Conseil a rejeté les demandes à ce qu'un accès désagrégé aux têtes de ligne soit accordé. Nous obliger à maintenir des points d'interconnexion désagrégés viendrait renverser cette décision, en plus de réintroduire l'asymétrie.
5773 Distributel a demandé que le Conseil mette un certain nombre d'options à la disposition des grossistes dans le but de faciliter la transition des points d'interconnexion existants vers le point d'interconnexion agrégé.
5774 Les câblodistributeurs sont disposés à maintenir en place des points d'interconnexion désagrégés pour les services existants, à tout le moins jusqu'à ce que les tarifs pour les points d'interconnexion agrégés soient finalisés, et à développer un accommodement raisonnable afin de répondre aux préoccupations relatives aux conditions contractuelles traitant des installations utilisées pour desservir les points d'interconnexion désagrégés.
5775 Merci de votre attention. Nous sommes maintenant prêts à répondre à vos questions.
5776 THE CHAIRPERSON: Thank you very much.
5777 Mr. Engelhart, can you walk me through your Annex 1? You say this shows that the disaggregated points are different than the POIs. I'm looking at it and I don't quite see it. Maybe you can explain it to me.
5778 MR. ENGELHART: Thank you, Mr. Chair. I will start, and I am going to ask my colleague, Dermot, if he wants to add anything,
5779 But what you see there is the plot of the different nodes in our network.
5780 THE CHAIRPERSON: Yes.
5781 MR. ENGELHART: And the time of day at which they achieve their peak usage in a given day.
5782 You will see that although what you could call the network peak happens at 11 o'clock at night, there is lots that have their peak at 10 o'clock at night, nine o'clock at night, eight o'clock at night, and there is even some that have their peaks at five in the morning or eight in the morning.
5783 So the idea that the network peak, which is the 11 o'clock at night, captures the peaks throughout the whole network is just plain wrong. The peaks occur at different times of the day in different nodes.
5784 THE CHAIRPERSON: I guess I don't understand the chart then. At the bottom you have the hours of the day.
5785 MR. ENGELHART: Yes.
5786 THE CHAIRPERSON: And you have two curves which I gather represent two different nodes. Is that the idea?
5787 MR. O'CARROLL: Commissioner, the two charts represent the two technologies in our network DOCSIS 2.0 and DOCSIS 3.0. DOCSIS 3.0 are used for much higher speed services.
5788 THE CHAIRPERSON: Right. And I see --so it's 12 per cent usage at one o'clock in the morning.
5789 MR. ENGELHART: Right.
5790 THE CHAIRPERSON: It seems to go down to basically zero at six o'clock in the morning.
5791 MR. ENGELHART: Yes. So at six o'clock there are actually are about 1 percent of the nodes that have their peak then and you are quite right, at one o'clock in the morning a little around 11 percent of the nodes have their peak at two in the morning. It will be about 7 percent.
5792 So this shows you the percentage of nodes that have their peak at those different hours of the day.
5793 THE CHAIRPERSON: Okay. So I should write on the left-hand side -- oh, on the vertical bar, this is the number of nodes that have peak -- that have their peak. That's what these various dots represent?
5794 MR. O'CARROLL: Commissioner, it's the percentage of the nodes that show their peaks. This chart is not a measure of usage in any node. It's the percentage of nodes at any point in the network which have their peak at that time of day.
5795 THE CHAIRPERSON: Okay.
5796 MR. O'CARROLL: And what we are trying to demonstrate is that different nodes have their peaks at different times based on the behaviour of the customer in that area.
5797 THE CHAIRPERSON: You say in paragraph 7:
"The information demonstrates that for Rogers' network, for example, the aggregate network peak does not correlate well at all with the peaks at the individual nodes."
5798 How does the chart show me that?
5799 MR. ENGELHART: So this is one day but if you pretended that that day, for example, was the day of the month that CNOC measured as the 95th peak, 95th percentile peak, they would take their five minute interval probably sometime around 11 o'clock at night because that's when the busiest point would occur at their point of interconnection but they would really only have captured for that day the peak behaviour at 13 percent of the network node. They would not have captured the peak behaviour in the other nodes.
5800 So it's sort of back to the analogy of the side streets. They are capturing the requirement for augmentation only at 13 percent of the nodes. They are not capturing the requirement for augmentation at the other 87 percent of the nodes by looking at that peak -- network peak.
5801 THE CHAIRPERSON: I am missing something. I'm sorry. I'm slow on this. I'm trying to get my head around it.
5802 If 13 percent have it at 11 o'clock and the others have it at a lower point, so why would they not be captured?
5803 MR. ENGELHART: It's not at a lower point. It's at a different time of the day. So if you can think of a network of roads Main Street (sic) has a peak at two in the afternoon, Elm Street has a peak at five in the afternoon and Main Street -- Main Street has a peak at 11 at night, so Maple at two, Elm at five and Main at 11.
5804 So CNOC would say, "Okay, look, we have measured the overall network peak and because Main Street is a bigger street that peak is at 11 o'clock at night so we have now effectively captured everything that's going on all of those roads".
5805 No, they haven't. By measuring the speed at 11 o'clock at night they are not capturing what the augmentation that Dermot has to do on Elm Street at a different time of the day and they are not capturing the augmentation that Dermot has to do on Maple Street at a different time of day.
5806 So that network peak is a very poor proxy. It only poorly correlates with the different peaks that drive augmentation throughout the network.
5807 THE CHAIRPERSON: Yes. You made that point before. I understood that. I just don't understand how this chart demonstrates it.
5808 MR. O'CARROLL: Maybe if I can try, Commissioner.
5809 The chart does not show the peaks at 11 or higher than the peaks at four. If you have a node that is peaking at 11 o'clock and a node that is peaking at four in the afternoon, the peak in those two nodes can be the same. It's just that there is different numbers of nodes at the different times.
5810 So a node with its peak at two or four in the afternoon can reach exhaust -- will reach exhaust at two or four in the afternoon. And as a result of that traffic we will provide capacity relief to that node. The traffic in that node at 11 may be much lower than the traffic at four in the afternoon.
5811 Other nodes have their peak traffic at 11. It's to do with the customer behaviour in that neighbourhood and the type of customer.
5812 MR. ENGELHART: All right. So I think what I --
5813 THE CHAIRPERSON: I mean let me --
5814 MR. ENGELHART: Yeah, sure.
5815 THE CHAIRPERSON: -- just finish this. I have got to get this into my head.
5816 Take your analogy of Main Street, et cetera and you measuring only that one or the point of interconnection, whichever you want. You are measuring the peak traffic there at 95th percentile. Why doesn't it automatically take care of all the others?
5817 That's what I don't -- you are measuring it there and that traffic will find its way onto the side streets, to your analogy, and will cause congestion there, no question about it. But it is captured at that point in time.
5818 You are trying to capture the augmentation costs now. You will know what the augmentations are at the point of interconnection but from that the model works with statistical averages presumably on what this will be on the other nodes, the downstream nodes. So where is there a cost here that's not being captured?
5819 MR. ENGELHART: So, say that on Elm Street at a given time of the day there was no traffic.
5820 THE CHAIRPERSON: Yes.
5821 MR. ENGELHART: Say that on Maple Street there was lots of traffic and say that on Main Street there was a medium amount of traffic, at the POI there would be an averaging of zero, a lot and medium and it would look like the network peak showed a medium amount of traffic. But, in fact, on Elm Street that is requiring augmentation.
5822 And a completely different time of the day Maple Street might have a lot of traffic and it would require augmentation and that wouldn't be captured by the network peak number at all.
5823 A completely different time of the day, Maple Street might have a lot of traffic and it would require augmentation and that wouldn't be captured by the network peak number at all.
5824 So the point is that since the traffic from those side streets reaches congestion or a maximum capacity at different times of the day and because there is an averaging of all those different streets by the time it gets to the freeway, you are simply not capturing the traffic that drives augmentation by capturing an average number at a single point in time.
5825 THE CHAIRPERSON: Let me make sure I understand your analogy. I thought by measuring it at a single point of interconnection we in effect measure it on the off-ramp from the freeway.
5826 So you are measuring all the traffic. You just don't know how it's being distributed over the system. Is that correct?
5827 MR. ROY PORRETTA: I'll take the highway example that was used earlier and multiple off-ramps. It's a small province. So there is one highway and there is multiple cities with different off-ramps.
5828 In the morning, people wake up in one city and so we take as a hypothesis at the highway that is basically a POI is built to sustain capacity in busy hours. So that's done. There is x-number of links. So in the morning people wake up in one village or one city. They go on the highway.
5829 Now, in the Internet business you won't accept congestion because if you have congestion basically people will not use the service or people will basically go somewhere else.
5830 THE CHAIRPERSON: Right.
5831 MR. ROY PORRETTA: So in the morning there is a certain capacity required in a certain off-ramp for that city. It's busy hour.
5832 So what we do is we measure this off-ramp 24 hours a day. It happens that the peak is there in the morning at eight. So we say, "Oh, there is a potential of congestion. We will build the off-ramp bigger". People flow from the morning from one city to another city in another off-ramp. So we build two off-ramps at eight a.m. in the morning and the highway is able to sustain everything.
5833 At five p.m. in the afternoon we will re-use the same off-ramp but if there are other peaks during the day where people need to move from one city to another city every off-ramp will be measured at their peak hour that can be any time of the day. It could be between noon. So there is maybe one place where people will go and eat between 12 and 1 p.m. because it's a great city for restaurants and people will move from one off-ramp to the other.
5834 So what you are saying is this highway is built and that's the POI. But every off-ramp depends on the time of the day and because we measure every time of the day because the peaks are at different times, we will adjust these.
5835 So if you take the hypothesis that everything is built by looking at the highway, you are not taking a good measurement.
5836 THE CHAIRPERSON: Okay. Taking the image you just now planted in my mind, then the CNOC model, the 95th percentile you say would only measure the traffic on the highway. It would not measure the traffic on the ramps. Is that --
5837 MR. ROY PORRETTA: Exactly.
5838 THE CHAIRPERSON: And that's your problem.
5839 MR. ROY PORRETTA: So it would measure the traffic on the highway but if there is a huge peak in one of the off-ramps -- and the example that you see in the graphic that Rogers has provided --
5840 THE CHAIRPERSON: M'hmm.
5841 MR. ROY PORRETTA: -- you will say, "Okay, this is the way all of the off-ramps are built". But it's not really true because depending on the hour of the day we will need build additional off-ramps because the Internet will not accept congestion.
5842 We are not in -- we accept congestion on the highway because we have to live with it, but on the Internet it's not something that people will accept.
5843 THE CHAIRPERSON: Okay. Thank you.
5844 MR. ROY PORRETTA: Is it clearer?
5845 THE CHAIRPERSON: I still don't understand how this chart relates, but I understood your point.
5846 MR. ENGELHART: Well, let me try this. If CNOC was right --
5847 THE CHAIRPERSON: Yes.
5848 MR. ENGELHART: -- this chart would look different. You would have at, say, 11 o'clock at night a single straight line that went up to 100 percent and at all other times of the day there would be no lines. So if they were right that the network peak captured the peaks throughout the network, you would just have a single vertical line at 11 o'clock and zero everywhere else.
5849 The fact that the graph doesn't look like that is evidence that they are wrong.
5850 THE CHAIRPERSON: Okay, thank you.
5851 Tom, you have some questions?
5852 CONSEILLER PENTEFOUNTAS : J'ai l'impression de tourner en rond, puis il n'y a rien de plus déraisonnable que d'avoir cette impression-là. Donc, on va essayer d'aller de point A à point B.
5853 D'abord, sur la facturation comme telle de la capacité maximale d'utilisation, si on n'était pas pour mesurer cette capacité sur le réseau comme un ensemble, y a-t-il d'autres endroits où on peut le mesurer?
5854 Disons qu'on tombe en amour fou avec le modèle de MTS Allstream, y a-t-il une autre façon de mesurer la capacité qui puisse être plus agréable ou plus raisonnable pour votre organisation, Monsieur Béland?
5855 M. BÉLAND : Ce qu'on a mentionné dans nos commentaires ce matin, c'est, en théorie, oui, on va mesurer la capacité maximale, le peak, à chaque noud du réseau. En théorie, c'est la perfection, sauf que, comme on a dit dans notre présentation, la perfection n'est pas sur la table.
5856 En termes pratiques, ça serait...
5857 CONSEILLER PENTEFOUNTAS : C'est pratiquement impossible.
5858 M. BÉLAND : ...ingérable. C'est pour ça qu'il faut chercher un proxy, et on a évalué en profondeur les deux proxies possibles qui semblaient être devant le Conseil, un proxy qui présume que l'utilisation maximale du réseau entier est bien corrélée avec les peaks, les sommets de chaque noud, ou un autre proxy qui présume que l'utilisation dans chaque noud est bien corrélée avec le sommet dans le noud.
5859 Et ce qu'on trouve dans notre annexe 1 et 2, c'est une analyse quantitative des deux proxies potentiels, puis on trouve que le deuxième est nettement meilleur en termes de sa corrélation factuelle avec les dépenses d'augmentation de notre réseau.
5860 CONSEILLER PENTEFOUNTAS : Est-ce pratiquement impossible de mesurer les sommets par noud?
5861 M. PORRETTA : Pour prendre l'exemple de l'analogie de " off ramp ", vous avez des voitures de différentes couleurs et de différents modèles.
5862 Alors, on regarde à une sortie qui a été bâtie en fonction du trafic peak, tous ces différents modèles de voiture, et on essaie de compter. Il y a une voiture de type A, 12 voitures de type B, 14 voitures de type C, dépendant du nombre de différents fournisseurs d'accès Internet que nous avons. De mesurer le trafic à un noud pour différents types de compagnies ou donc différents types de voitures devient extrêmement difficile.
5863 Alors, ce qu'on a trouvé, c'est le proxy qui est le plus simple, qui dit nous allons mesurer le volume, donc la quantité de voitures qui vont passer dans le temps vers chacune des différentes destinations, et cette mesure-là va nous permettre d'arriver -- et les corrélations le démontrent -- à une mesure qui représente la réalité.
5864 Comme on dit, si c'était parfait, on aurait plusieurs milliers de nouds à aller mesurer, un par un. Imaginez-vous quelqu'un cligne des yeux, parce que la personne est payée pour mesurer ces différentes voitures-là, on a un gros problème.
5865 Alors, notre mesure, actuellement, on considère -- et les corrélations l'indiquent -- que c'est une très bonne représentation de l'utilisation peak, et elle est mesurée à chacun des nouds plutôt que d'être mesurée sur un endroit que, anyway, a été bâti pour prendre le peak.
5866 CONSEILLER PENTEFOUNTAS : Je comprends.
5867 Si on regarde également l'alinéa 18, sur les coûts de l'application du système comme tel, ça risque d'être onéreux, ça risque de coûter une fortune d'implanter ce système-là.
5868 Si CNOC et les autres étaient prêts à assumer ces coûts-là, ça ne poserait pas de problème pour votre groupe?
5869 M. BÉLAND : Je crois que Ken voudrait commenter plus en profondeur, mais nous, notre critique principale du modèle 95e, que ça soit la saveur CNOC ou la saveur MTS, notre critique principale et dominante, c'est le fait que le modèle ne reflète pas les coûts de notre réseau et que le modèle créerait des incitatifs pour changer les coûts de notre réseau.
5870 Donc, les coûts d'implantation, c'est une considération importante pour nous. C'est comme une sorte de gaspillage. On a deux modèles possibles. Il y en a un qui peut être implanté facilement parce qu'on mesure déjà les volumes, un autre modèle qui exige des investissements et qui arrive en fin de compte à avoir une pire performance.
5871 Mais ceci étant dit, bien sûr, si le modèle 95e était adopté, bien sûr que les coûts seraient supportés par les fournisseurs des entreprises de gros.
5872 CONSEILLER PENTEFOUNTAS : Je comprends. Ça sera un problème de moins. On peut s'entendre là-dessus, Monsieur Béland?
5873 M. PORRETTA : J'aimerais juste rajouter un peu.
5874 Nous mesurons déjà à tous nos équipements par modem. Là, je prends vraiment un volet technologique, ingénieur. Nous mesurons, pour des millions de modems à travers tous les cablôdistributeurs, le volume. La façon que ça fonctionne, c'est qu'on a des mesures qui sont dans nos équipements, et on ne demande pas une réponse de la mesure à toutes les cinq minutes. Ce qu'on fait, c'est qu'on extrait l'information à certains cycles. Disons que c'est aux demi-heures. C'est des millions d'information qu'on extrait de temps à autre.
5875 Dans le cas où on allait -- prenons le cas de Vidéotron, 8 500 nouds -- on allait extraire dans 8 500 nouds de l'information sur X nombre de fournisseurs Internet à toutes les cinq minutes. C'est le type d'information qu'on doit mesurer dans nos systèmes et extraire. Ça devient extrêmement difficile. De mesurer le 95e percentile 8 000 fois, c'est une chose, mais de le mesurer 8 000 fois toute la quantité de divers fournisseurs et la complexité devient extrêmement élevée.
5876 L'autre chose qu'on a déjà discutée dans une autre instance, c'est qu'on ne veut pas se retrouver avec un équipement qui n'est bâti que pour le Canada. On veut pouvoir réutiliser -- et ça, c'est extrêmement important -- des équipements qui sont bâtis de façon à avoir une amélioration dans notre volume.
5877 Si on demande des spécifications pour le Canada seulement, ça veut dire qu'on va dire, mais vous, dans cet équipement-là, ce n'est pas une mesure de volume, mais c'est une mesure précise par fournisseur par cinq minutes à travers tout le réseau. On va leur dire : Bien, mettez plus de mémoire, mettez plus de processus. On veut que le (indiscernable) soit plus performant.
5878 On vient encore de prendre un chemin qui nous limite dans notre choix d'équipement au Canada. Déjà, on a certains impacts avec ce qu'on est en train de faire présentement, mais on va accentuer encore plus le volet, les coûts et la complexité de faire ça.
5879 CONSEILLER PENTEFOUNTAS : Alors, pratiquement parlant, c'est impossible, si je vous comprends bien.
5880 Vous avez également soulevé un autre point sur l'internationalisation, si vous voulez, qui n'a pas été touchée trop, trop jusqu'ici.
5881 Les us et coutumes se basent sur la facturation par volume et non pas par capacité; c'est également...
5882 M. PORRETTA : Oui, c'est également vrai, tous les cablôdistributeurs au Canada et aux États-Unis se basent sur le volume en ce moment, effectivement.
5883 CONSEILLER PENTEFOUNTAS : Et ce serait un désavantage considérable de vous soustraire de ce qui est la pratique à travers le continent?
5884 M. PORRETTA : Ou de nous imposer une pratique différente. Ce serait effectivement nous donner des efforts supplémentaires, des coûts, de la complexité, et ça va affecter notre capacité canadienne, chez les cablôdistributeurs, d'innovation, et ça, c'est extrêmement important pour nous.
5885 CONSEILLER PENTEFOUNTAS : Vous avez soulevé une autre question -- je pense à l'alinéa 31 -- où vous parlez du fait que les gens vont vouloir maximiser ce qu'ils ont acheté, et ça revient un petit peu dans le thème qui est connu comme le gaming comme tel. On va essayer de maximiser notre utilisation du fait qu'on a payé pour cette capacité maximale.
5886 Il y avait un débat la semaine passée : Est-ce qu'on est vraiment capable d'essayer d'atteindre ce maximum? Est-ce techniquement et pratiquement faisable?
5887 Pensez-vous que c'est le cas, que les gens puissent, une fois qu'ils ont atteint cette capacité puis ils sont pris en quelque sorte de la payer, qu'ils vont pouvoir rester très, très proches de cette capacité maximale?
5888 M. BÉLAND : Notre préoccupation, ce n'est pas autant qu'ils vont nécessairement atteindre le sommet 24 heures par jour avec perfection. La préoccupation, c'est que vous allez... si vous adoptez cette méthode de facturation, vous allez créer des incitatifs pour le faire, pour chercher des façons de le faire, et...
5889 CONSEILLER PENTEFOUNTAS : Mais est-ce faisable? Je comprends que l'incitatif est là, mais est-ce pratiquement faisable?
5890 M. PORRETTA : Je vais demander à Ken d'ajouter son point de vue.
5891 CONSEILLER PENTEFOUNTAS : Si c'est une préoccupation légitime, c'est ça la question.
5892 M. PORRETTA : Oui. Je vais demander à Ken.
5893 CONSEILLER PENTEFOUNTAS : Merci.
5894 MR. ENGELHART: Yes. We gave one example in our in-chief remarks when we spoke last week, which would be a service, say, for business customers that operated from 9:00 to 5:00.
5895 We heard another example today, which is the ISPs, the wholesalers, could do usage-based billing but free usage from 2:00 in the morning until 6:00 in the afternoon.
5896 The reason why that is quite feasible in the Internet world is because a lot of the traffic is machine-to-machine. So in a human being network, they tend to go to sleep at night, they tend to be awake during the day. So in the telephone network that a lot of us think about when we think about these capacity issues, there's a natural trough at night.
5897 What we're finding in the Internet world is the trough is still pretty close to the peak and we're finding that even in the wireless network now too because people are asleep but their iPads and their iPhones are working away all night. And so you can program your computer to do its thing at night or at 2:00 in the morning or whenever you want. So shifting those peaks is quite easy.
5898 COMMISSIONER PENTEFOUNTAS: Thank you, Mr. Engelhart. I think my other colleagues may have some questions.
5899 THE CHAIRPERSON: Now, your whole argument that you just made before and that Bell made is that you're saying is our cost is based on a certain percentage of usage. You know, we try to estimate it, et cetera, and if you adopted the MTS model, then they have an incentive to drive everybody to 100 percent and then our costing is wrong because it's based on average use rather than maximal use.
5900 What you just said seems to imply that that's yesterday, nowadays people are actually using things to their maximum even while they're sleeping.
5901 MR. ENGELHART: No. Because we charge our own customers and because we charge -- we're proposing to charge the wholesalers per gigabyte, then people will use those gigabytes when it's most convenient for them to do so.
5902 But if you move to this peak-capacity model, you create an incentive for them to flatten those peaks and that will create a distribution which our costing models are not based on, as you said.
5903 THE CHAIRPERSON: I wasn't talking about 95th percentile. I was talking about, let's say, the Primus fallback model or the MTS model which basically -- both you and Bell said that will encourage people to push their users up to the absolute maximum and our costing model isn't based on that. It's not based on people using it to their maximum but using it sort of in an average way that you somehow calculate.
5904 But you just said, you know, we have to revise our thinking because people are using the network more and more even in the off hours and there are no more peaks.
5905 Wouldn't the logical conclusion be that the MTS model is actually an acceptable model?
5906 MR. ENGELHART: I don't think I made that last statement that there are no more peaks. It's not as "peaky" as it used to be and the troughs are not as low as they used to be. There are still peaks, there are still troughs, but, no question about it, it's getting flatter in a machine-to-machine world than it was in the old voice world.
5907 THE CHAIRPERSON: Right. Okay.
5908 MR. BÉLAND: Mr. Chair?
5909 THE CHAIRPERSON: Yes.
5910 MR. BÉLAND: I think Mr. Jetté would like to add something.
5911 M. JETTÉ : Si je pouvais ajouter sur la probabilité que les changements peuvent être faits au réseau aujourd'hui.
5912 Il faudrait regarder les composantes qui constituent le service d'accès Internet, et en plus du nombre d'usagers, il y a la vitesse d'accès au réseau, la fréquence d'utilisation d'un usager sur ce réseau-là et la durée qu'un usager peut utiliser le réseau.
5913 Les réseaux, l'ingénierie des réseaux est déterminée selon la probabilité d'atteindre certaines vitesses, d'avoir certaines fréquences d'utilisation au réseau ou de rester en connexion permanente sur le réseau.
5914 Donc, c'est entièrement possible qu'ils peuvent modifier les paramètres de vitesse, les paramètres de fréquence d'utilisation et les paramètres de durée d'utilisation pour créer des nouveaux modèles de produits sur le marché et changer l'utilisation de nos réseaux ou la capacité simultanée utilisée sur nos réseaux.
5915 LE PRÉSIDENT : O.K. Michel?
5916 CONSEILLER MORIN : Merci, Monsieur le Président.
5917 D'abord, je vous remercie d'avoir déposé des chiffres en réponse à ceux que j'avais moi-même déposés. Je pense que le personnel va sûrement trouver tout ce qu'il faut là-dedans pour éclairer le Conseil sur la décision à prendre.
5918 Et justement, si le Conseil devait éventuellement choisir le modèle de MTS, ce matin, les gens de CNOC nous ont dit qu'ils pourraient, à tous les trois mois, offrir en quelque sorte une prévision de la demande.
5919 Est-ce que ce serait pour vous quelque chose qu'il faudrait retenir comme proposition?
5920 M. BÉLAND : Le commentaire que je ferais, c'est -- et je crois que c'est fondamental -- la commande, l'approvisionnement des installations d'interconnexion, les installations physiques d'interconnexion, c'est quelque chose qui fonctionne déjà. Ça fonctionne bien aujourd'hui.
5921 Vidéotron... J'ai donné l'exemple de l'ordonnance récente du Conseil concernant les installations gigabits Ethernet de Vidéotron. On n'a pas de problème avec... de pouvoir s'asseoir avec les entreprises de gros, d'évaluer les besoins, d'évaluer les tendances et de provisionner les installations qui sont nécessaires.
5922 Je crois que le modèle MTS, plutôt que de donner plus de succès à ces échanges-là, va plutôt créer des problèmes, puis la raison, c'est la suivante.
5923 En prenant le point d'interconnexion, en prenant l'installation physique au point d'interconnexion et en décidant c'est là où on va rajouter plein de frais, c'est là où on va chercher de récupérer le coût du réseau qui est en arrière, le réseau d'accès -- et ce sont des frais substantiels qu'il faudrait rajouter, c'est sûr -- là, on va créer des conflits, parce que là, on va voir des entreprises de gros qui disent, ah, on provisionne en trop, j'aimerais des incréments qui sont plus petits, je trouve que je paie trop puis je n'ai pas l'utilisation suffisante.
5924 Ce genre de conflit, dans le cas de Vidéotron, n'existe pas aujourd'hui, et j'ai peur que vous allez créer plus de conflits de cette nature-là si vous misez trop sur mettons le fardeau de la récupération des coûts sur l'installation physique au point... à l'interface.
5925 CONSEILLER MORIN : C'est dans ce sens-là, justement, que je posais la question.
5926 Ce matin, au paragraphe 8, CNOC a bien dit que l'approche "providing quarterly forecasts of peak demand."
5927 Alors, dans ce sens-là, si toujours le Conseil devait adopter le modèle MTS, est-ce qu'il ne faudrait pas s'assurer que vous n'êtes pas pris par surprise par les demandes des petits fournisseurs Internet, qui, vous m'avez cité ce matin en disant que je vous invitais à prendre vos responsabilités, mais le modèle MTS, soit-dit en passant, invite aussi les petits à prendre leurs responsabilités, à prévoir leur demande, à évaluer leurs besoins, à choisir et à prendre des risques d'affaires.
5928 Alors, dans tout ce grand univers, la question que je vous posais : Compte tenu qu'on serait dans un nouveau mode, est-ce qu'il ne serait pas essentiel que la Commission exige que dans ce modèle-là, s'il devait être adapté toujours, eh bien, il y a une prévision qui vous soit faite à tous les trois mois de leurs besoins éventuels et possibles?
5929 M. BÉLAND : Il n'y a aucun fournisseur de service de gros qui va refuser un client qui veut donner des prévisions, c'est sûr.
5930 CONSEILLER MORIN : Donc, vous avez beaucoup de capacité...
5931 M. BÉLAND : C'est sûr. Si un client veut nous donner des prévisions d'avance de X mois, il n'y aucun fournisseur de service qui va le refuser.
5932 Le problème qu'on a, encore une fois, c'est de tous les risques que nous, on coure dans cet exercice, un des moindres, c'est notre capacité de s'asseoir et de prévoir avec une entreprise de gros ses besoins en installation physique au point d'interconnexion.
5933 Le risque qu'on coure dans cet exercice -- et, encore une fois, je m'excuse de le répéter -- c'est votre modèle de facturation que vous allez adopter, est-ce que ça va créer des incitatifs qui vont changer la structure de coûts de notre réseau qui est en arrière. Notre focus est dessus.
5934 CONSEILLER MORIN : Vendredi, j'ai posé la question à Vaxination : Oui, mais si on prenait le modèle MTS, peut-être ça va très bien avec les autres telcos, mais avec les câblos, est-ce que ça va aussi bien?
5935 Et le consultant nous a dit : Bien, il n'y a aucun problème, selon moi, qui soit... ce modèle-là pourrait être appliqué aussi bien aux entreprises de câble, aux cablôdistributeurs, qu'aux entreprises de téléphone.
5936 Est-ce que vous êtes de cet avis-là?
5937 M. BÉLAND : Dans une perspective purement technique, est-ce qu'il est possible d'implanter le modèle? La réponse est oui. Mais, encore une fois, je répète, ça ne répond pas à nos préoccupations avec le modèle.
5938 CONSEILLER MORIN : Merci.
5939 Ce sont mes questions, Monsieur le Président.
5940 THE CHAIRPERSON: Candice...?
5941 COMMISSIONER MOLNAR: Thank you.
5942 First of all I want to go to your comments about the access rate and whether it should be a single access rate or whether it should vary by speed.
5943 I think I understand that you don't support a single access rate, but you do support, do you, a rate that is cost-based?
5944 MR. BÉLAND: Yes, we do. The entirety of the rates that we have put forward we support that they be cost-based and they have been derived on cost-based principles.
5945 COMMISSIONER MOLNAR: Right. So to the extent that the costs change with speeds, if those costs are reflected in the rate, you are comfortable with that outcome?
5946 MR. BÉLAND: Absolutely. In fact, that's exactly what we are saying. We are saying that the costs of access do change with speed, they change substantially with speed, and if that reality is reflected in the rate structure then I think we are perfectly in line.
5947 COMMISSIONER MOLNAR: Okay. I just want to make sure I do really fully understand this, because CNOC spoke about blending the usage charges that were in the access with the usage related to the aggregation portions of the network I believe to get a usage charge. If I understood them right, their usage charge would take some of the usage associated with the different speeds and access and blend it with the aggregation.
5948 Is that what you understood from their presentation this morning? And use some kind of mark-up that would be reflective of the different portions of the network?
5949 MR. BÉLAND: That's not quite what I understood from their presentation.
5950 Maybe we can ask them, but I think the point that I heard and the point that we would contest vigorously is the assertion that the cost of the access is independent of speed.
5951 If you do the exercise -- which we think is rather abstract, but if you do it -- of stripping out usage completely and look only at access, we definitely do not agree with the assertion that the cost of the access is independent of speed.
5952 I could ask some of our engineering experts to go into more detail if you would like.
5953 COMMISSIONER MOLNAR: I don't think that's necessary. You have filed cost studies that support the different costs associated with your different speeds, so I think we are okay there.
5954 But I wouldn't mind, if I could, having CNOC just clarify this point in case I misunderstood what they have presented.
5955 MR. TACIT: Commissioner Molnar, it's Chris Tacit.
5956 Your interpretation is correct and we have also said that to the extent that there are non-traffic sensitive costs that do vary by speed, that would be a proper basis for having different speed rates.
5957 We only said that where cost do not differ there should be no difference in the rate.
5958 COMMISSIONER MOLNAR: Okay. Thank you.
5959 THE CHAIRPERSON: Do we have a rare agreement between CNOC and Rogers and the cable companies here?
5960 THE CHAIRPERSON: Okay, I just wondered.
5961 Fine. Let's go on.
5962 COMMISSIONER MOLNAR: Thank you.
5963 I do have another question and it relates to your Schedule 1. I just want to make sure I understand.
5964 First of all, can you just confirm for me that this is your entire inventory -- this is Rogers -- this is your entire inventory of CMTS.
5965 So would this include both residence and business?
5966 MR. ENGELHART: Yes.
5967 COMMISSIONER MOLNAR: And also, just again to make sure I'm understanding this schedule, this is where the traffic peaks, at what time the traffic peaks, but this is not, if I understand correctly, the time at which you may need to augment.
5968 Is that correct?
5969 MR. O'CARROLL: Commissioner, what this shows is, it does show the time at which we would augment individual nodes, because the traffic peaks for that node at that time. We monitor --
5970 COMMISSIONER MOLNAR: Does that mean that the CMTS is at exhaust at that time or just that's when the traffic peaks, for that node that is the highest traffic peak or is that when you have to augment that node because the node has peaked?
5971 MR. O'CARROLL: So this is showing the state of the traffic when we measured it.
5972 So those nodes would not have been at exhaust at this point, but the traffic through the day at that node would have peaked at that point. The probability is at some point when they do need augmentation it would be at an hour such as this during the day.
5973 So the exhausts do happen throughout the day in a manner similar to this.
5974 COMMISSIONER MOLNAR: Do you have a graph that would show us that, that would relate when you augment to the time of day in which it peaked? Because this just says this is when it has the highest amount of traffic for both your residence and business, it doesn't say this is at the point at which you have made an investment into these nodes.
5975 MR. O'CARROLL: So the way we make investment is when that peak reaches 60 percent of the capacity we start planning scheduling relief. When the peak reaches 70 -- so that we provide capacity relief prior to the capacity reaching 70 percent.
5976 So we never actually -- try never to allow a node to reach peak capacity or to exhaust and those peaks -- the timing of that would logically be similar to the timing on this chart.
5977 COMMISSIONER MOLNAR: Okay. But just confirm for me, is it possible that these nodes that you are showing that are peaking at, you know, 1:00 in the morning -- or not 1:00 in the morning --
5978 MR. O'CARROLL: Some of them at 1:00 in the morning.
5979 COMMISSIONER MOLNAR: Yes, some at 1:00, but these that are peaking at, you know, 5:00 in the morning, 6:00 in the morning, may never actually reach exhaust.
5980 MR. O'CARROLL: Oh no, no. No, they all reach exhaust.
5981 COMMISSIONER MOLNAR: They all will reach exhaust?
5982 MR. O'CARROLL: Oh yes, absolutely.
5983 I mean the traffic -- I mean the chart can look misleading because the chart -- the graph is low at 5:00 doesn't mean the traffic is low. Traffic on these nodes all reach the -- I mean the average node will last -- at 40 percent growth per annum the average node only last two years. So on average we relieve every single node on our network every two years. So they all reach exhaust, absolutely.
5984 We are doing capacity relief on these nodes, multiple nodes, every single night of the week every week of the year to keep up with the capacity and growth.
5985 COMMISSIONER MOLNAR: Okay. I guess not to -- I'm just trying to understand --
5986 MR. O'CARROLL: Yes.
5987 COMMISSIONER MOLNAR: -- because to me when I look at this I look at there are some nodes whose peak traffic is at some very surprising times of the day.
5988 MR. O'CARROLL: Yes.
5989 COMMISSIONER MOLNAR: I mean they could be running business, they could be transporting.
5990 MR. O'CARROLL: Well, I mean they could be -- as Ken mentioned earlier, a lot of traffic is machine-to-machine, which is typically peer-to-peer traffic. That traffic tends to be steady stream continuous traffic, which is much more consumptive of capacity than peak human usage.
5991 COMMISSIONER MOLNAR: Yes.
5992 MR. O'CARROLL: So a relatively small number of steady streams will cause exhaust in a node and they happen at 4:00 in the morning. That node exhausts at 4:00 in the morning.
5993 COMMISSIONER MOLNAR: And there is nothing you have that would show us where you would be able to relate when you actually are augmenting or replacing these nodes?
5994 MR. O'CARROLL: I would say that this chart would represent that --
5995 COMMISSIONER MOLNAR: It does.
5996 MR. O'CARROLL: -- very accurately. Absolutely.
5997 COMMISSIONER MOLNAR: It represents your investments.
5998 MR. O'CARROLL: Absolutely. The timing at which node exhaust causes our investment.
5999 COMMISSIONER MOLNAR: Okay.
6000 MR. O'CARROLL: This chart represents that.
6001 COMMISSIONER MOLNAR: Okay, thank you.
6002 While you don't have a residence only view, do you have a sense as to how this picture would look if you had a residence only view?
6003 MR. O'CARROLL: I'm sorry, I don't have that.
6004 COMMISSIONER MOLNAR: Okay. Thank you.
6005 THE CHAIRPERSON: Just follow this up, Mr. O'Carroll, I don't quite understand it.
6006 This chart that you gave us was obviously a certain period of time, I don't know when it was, but for that same period of time you are not telling me that every single one of these nodes you decided to augment. You, by yourself, say there is some sort of percentage, 60 or something, when you -- take that same period and superimpose on me which nodes you will have to augment because you are reaching 60 percent, or whatever your magic number is?
6007 Surely you can produce something like that for us.
6008 MR. O'CARROLL: Yes. This doesn't say that these nodes exhausted in this time period.
6009 THE CHAIRPERSON: Exactly. But that was --
6010 MR. O'CARROLL: Okay.
6011 THE CHAIRPERSON: My colleagues question was to give us -- maybe you want to file it in confidence, I'm not trying to get you to reveal your business, but you showed us that there is a different peak for different nodes --
6012 MR. O'CARROLL: Yes.
6013 THE CHAIRPERSON: -- but that doesn't mean that these nodes are close to exhaustion point and have to be --
6014 MR. O'CARROLL: That's right.
6015 THE CHAIRPERSON: So this period, for what you measured here, can you not also indicate which ones are at what you consider the exhaustion point and therefore have to be augmented?
6016 MR. O'CARROLL: What we would have to do is, at any point in time look at the utilization of all the nodes and identify at that point in time which nodes are at exhaust and record that and present it.
6017 I mean we could do that.
6018 THE CHAIRPERSON: You presented this chart to me which represents a certain time period.
6019 MR. O'CARROLL: Yes.
6020 THE CHAIRPERSON: To reinforce your argument it would be very helpful if on that same chart, first of all, you could show which ones of these are actually at exhaustion, therefore you --
6021 Second, the point that Mr. Engelhart made that this does not represent at all the anomalies per network, if you measure it by 95th percentile it would look totally different.
6022 MR. O'CARROLL: So this was taken at a period in the month of May so I would have to go back.
6023 THE CHAIRPERSON: Okay, take the month of May.
6024 MR. O'CARROLL: I would have to go back and see if we could pull out the records of the exhaust.
6025 THE CHAIRPERSON: Okay.
6026 MR. O'CAROLL: I will undertake to go and see if we could do that.
6027 THE CHAIRPERSON: If you could do that it would be very much appreciated.
6028 MR. O'CARROLL: We can do that.
6029 THE CHAIRPERSON: Okay. Len...?
6030 COMMISSIONER KATZ: Thank you, Mr. Chairman and good morning.
6031 I first of all thank you for filing the correlation statistics that you did. In paragraph 10, just so I can understand what you are saying, you are saying that peak is not correlated with usage, but usage is correlated with peak.
6032 Is that correct?
6033 MR. ENGELHART: No. The word "correlation" means that the two move together.
6034 COMMISSIONER KATZ: Yes. So the more usage, the more peak.
6035 MR. ENGELHART: Right.
6036 COMMISSIONER KATZ: But the more peak doesn't necessarily mean more usage?
6037 MR. ENGELHART: Well, what Schedule 2 shows is that for most of our nodes the more traffic on that node, the more peaky is the distribution at that node. So that is why we say that volume is a good proxy.
6038 COMMISSIONER KATZ: Yes, but the more peaky does not necessarily mean more usage, the converse?
6039 MR. ENGELHART: Well, 79 percent of the time it does.
6040 So if you look at the chart, if you would have a look at Schedule 2, for example way out on the right there is a little outlier dot. That is a strange node which has very high volume, but it's sort of flat. It's just not very peaky in that node. They are using tons and tons of volume, but it's just not very peaky at that node.
6041 If you look on the left-hand side of the chart there is a little outlier out there, too, and that shows on node with fairly low volume, but it's very peaky in that node. The little bit traffic that's being used is really all coming at a peak period.
6042 But if you look at that solid line, what that tells you is the more volume there is in a node -- I'm going to ask Dermot to jump in because he may have a different way of explaining it, but the more traffic there is at the node the peakier that traffic tends to be.
6043 MR. O'CARROLL: Commissioner, I think what we had said was that the peak at the POI correlates to usage on a subset of nodes, not on all of the nodes, but the usage at each node correlates very well with the peak of each node.
6044 COMMISSIONER KATZ: Right. So the regression analysis of .7874 reflects the correlation. If you ran it the other way and tried to regress peak against usage, would that be the 13 -- the R-squared would be 13 percent which you are referring to in paragraph 10?
6045 MR. ENGELHART: So if you were looking at an individual node, the correlation is 79 percent whether you are correlating peak with volume or volume with peak.
6046 What we did in Schedule 1 is, we correlated not to peak at the node with the volume at the node, what we are saying, if you look at the network peak there is only a node peak at 13 percent of the nodes. So that's where we are getting the 13 percent figure from, is from Schedule 1, and that's because you are now looking at an aggregate POI peak and trying to look at what is happening at the individual nodes at that time.
6047 COMMISSIONER KATZ: Yes, okay. So we have two different ways of calculating, a 13 percent and a 79 percent R-squared.
6048 What I'm asking is: Can you run Schedule 2 with the reverse regression and see whether the R-squared is higher than 78 percent or not, to confirm your hypothesis that there is a strong correlation with the more you use the more peak you are going to create, versus the more peaks there are, the more you use?
6049 MR. ENGELHART: It has been a long time, Commissioner Katz, since I studied statistics, but I don't think a correlation and a regression are the same thing even though they are similar.
6050 So with a correlation you are not regressing one variable against another variable, you are looking at the relationship between those two variables. So whether you regress volume on peak or peak on volume, it's the same R-squared.
6051 Mr. Watt studied statistics more recently than I did, maybe he can jump in.
6052 MR. WATT: Yes, it is the same correlation. It doesn't depend whether one is labelled the dependent variable and the other the independent variable, they are just the two variables here so the correlation is the same.
6053 COMMISSIONER KATZ: I graduated from statistics.
6054 COMMISSIONER KATZ: Okay.
6055 The other question I have is in paragraph 14. You talk about volumes doubling and tripling.
6056 If the wholesale regime or the independent ISPs have the flexibility that they are seeking -- and you are saying under that scenario there would be an imposition of substantial incremental costs in the underlying network that would be imposed upon you -- are you saying that purely because -- back to the issue of gaming -- there would be an incentive to try and drive usage, or are you saying that irrespective of that situation there would be increase in volumes of usage simply because they can?
6057 MR. ENGELHART: No, it's the gaming.
6058 COMMISSIONER KATZ: It's the gaming.
6059 MR. ENGELHART: It's incenting their end users to behave in a certain way.
6060 COMMISSIONER KATZ: Okay.
6061 My last question, paragraph 17. You say the 95 percentile calculation would result in higher costs of billing.
6062 Have you estimated what the impact on the costs would be and the timelines to do it?
6063 MR. BÉLAND: Mr. Vice Chair, Vidéotron, for one, has provided an estimate in confidence in one of our interrogatory responses. I believe it was in the last round of interrogatories.
6064 COMMISSIONER KATZ: Okay.
6065 How about Rogers and Cogeco?
6066 MR. MESSIER: Same thing for Cogeco.
6067 COMMISSIONER KATZ: So you filed that information in confidence?
6068 MR. WATT: No, we didn't file ours in confidence. We haven't actually been able to do a detailed study, we simply said that for every million dollars that was incurred it would relate to a certain cost per TPIA end-user. We actually did file that latter point in confidence.
6069 But I think it's important to point out that all the numbers that were filed related only to measuring at the aggregate POI point.
6070 We have absolutely no estimate of the complexities that Pierre spoke about earlier this morning in terms of calculating at each individual node, which is where you would want to capture the traffic if you are going to have the costs line up with the traffic that's actually going through.
6071 COMMISSIONER KATZ: Thank you.
6072 THE CHAIRPERSON: Okay. Those are our questions.
6073 One intervenor -- no, two, and then let's move on to Shaw. I would like to do Shaw before we break for lunch.
6074 So let's start with you, Bell. What did you have?
6075 MR. BIBIC: Mirko Bibic, for Bell.
6076 I was just going to add -- and we can cover this this afternoon hopefully, or tomorrow -- to the Vice Chairman's question.
6077 I think what Rogers Schedule 2 shows is that traffic is a good proxy for peak set nodes. That's what the Rogers Schedule 2 shows. So then from that we glean that peaks are caused by traffic volumes at the nodes.
6078 So the key question to us becomes: Do the peaks at the A-HSSPI, or at the point of interconnection between a network provider and the ISP, do peaks there correlate with volume in the aggregation network? If so, then maybe -- maybe 95th percentile would work.
6079 We will show this afternoon, because we have a chart, that there is no correlation between the peaks at the A-HSSPI and what's happening in the aggregation network.
6080 So the 95th is not a good proxy for what's happening in the aggregation network. So when we come forward this afternoon with our chart, which is a correlation with what's happening at the A-HSSPI, you can compare it to what's happening in the node. You put the two together and it will establish, in my submission, that 95th is not a good proxy for what's happening in the aggregation network.
6081 THE CHAIRPERSON: Okay.
6082 The gentleman in the far corner there.
6083 MR. MEZEI: Jean-François Mezei, Vaxination Informatique.
6084 In the Schedule 2 graphs from the cable companies, number of bytes is rather meaningless -- sorry to be blunt -- there should be a unit of time in there. Is it per month? Per week? Per second? Because if it's per second it's a unit commonly used, kilobytes per second. It's a unit of capacity.
6085 If you have essentially units of bytes per unit of time, it does reflect capacity to a certain extent.
6086 Now, if you have it over a full month it is averaged out. If you have it over a few seconds, then it's more of a point thing.
6087 In the case of 95 percentile, it measures gigabytes per second and it's very precise, whereas what the cable companies want is something that is brought over a period of months. And over a period of months you don't measure peaks that really affect what your network needs to be provisioned for, in other words, your capacity demands.
6088 THE CHAIRPERSON: Well, let's hear from the cable companies, what's the time period?
6089 MR. O'CARROLL: These actually represent the average per day usage. It's the average daily usage on each of the nodes
6090 THE CHAIRPERSON: His specific point, if I understood his point, was bytes.
6091 Are you doing it per second? Are you doing at per day?
6092 MR. O'CARROLL: It's bytes. It's bytes. It's the average cumulative bytes per day on each of the nodes.
6093 THE CHAIRPERSON: How do you answer to his point that this is meaningless?
6094 MR. O'CARROLL: No, I think it's the opposite. This is the average usage by the customer on a daily basis and that average usage correlates very well with the peak of the node.
6095 MR. BÉLAND: Mr. Chair, we are attempting to show a correlation between volume and peak, so clearly the axis that we are going to show that we are going to give you are volume and peak.
6096 It's no more complicated than that.
6097 THE CHAIRPERSON: Whoever is making this observation...?
6098 MR. COHEN: It's Mel Cohen.
6099 It's just a question. I don't understand the "Y" axis. I don't understand what it is. We just had the "X" axis explained to us, but can they please tell us what the "Y" axis is?
6100 THE CHAIRPERSON: What is the specific question you want the cable companies to answer?
6101 MR. COHEN: What the "Y" axis means for a single node that's graphed on the chart.
6102 MR. O'CARROLL: It simply represents the peak utilization of that node at the 95th percentile. So we are trying to show the correlation between volume of traffic in the node with peak utilization at the 95th percentile.
6103 THE CHAIRPERSON: Okay.
6104 Mr. Rocca...?
6105 MR. ROCCA: Peter Rocca.
6106 MR. BHULLAR: Sorry, Mr. Chairman, I think there is -- as you started on Schedule 1 there is obviously confusion and I think Rogers has successfully done so. I think lots of the people here are confused for sure.
6107 The question is, when they say there is a peak, first of all, if they provide relief at 60 percent they shouldn't be finding any nodes in the network at 95 percent. So that's one point.
6108 The way they have shown it here, it seems that either they are using the wrong assumptions about traffic engineering a node; or they have not adjusted the traffic engineering principles to the mix of traffic on that node; or third, they have not doubled home or better interconnected the nodes to load share if it is a transport congestion.
6109 So they should clarify whether it is a CPU congestion within the node or it is a transport construction, links coming out of the back end of the node around the ring that they have, say, around for Wolfdale or for 333 Bloor, they have to clarify what is the speed? Is it that the node can't process any more Internet traffic or that the links going out of the node are full? If the links going out of the node are full, that is simple provisioning, they have fibre running around and it is very little cost generally, right.
6110 But if it's a node exhausted --
6111 THE CHAIRPERSON: Okay. Okay, hold it. One question at a time. You have major question.
6112 Mr. O'Carroll...?
6113 MR. O'CARROLL: Mr. Chair, first of all what it shows is not that the nodes are at 95 percent of capacity and therefore they exhaust, it's the graph showing the 95 percentile point of the peak of that node. So if there was 20 megabits per second traffic at a point in that node, it would be 95 percent of bats. It's the 95th percentile of the peak of the node.
6114 With respect to the other comment, I'm not 100 percent sure what the person was getting at.
6115 THE CHAIRPERSON: I'm sorry, you lost me in your explanation. Can you simplify that?
6116 MR. O'CARROLL: This chart is trying to graph the peak traffic that happened at each node at the time of day at which peak traffic happened and is using the 95th percentile peak --
6117 THE CHAIRPERSON: Yes.
6118 MR. O'CARROLL: -- as the point in time. That's all it is.
6119 It's not indicating that the traffic in the node was at 95 percent of its capacity.
6120 THE CHAIRPERSON: Okay.
6121 MR. O'CARROLL: With respect to the other comment, I'm not sure the questioner understands our particular architecture.
6122 MR. BHULLAR: I can clarify it.
6123 MR. O'CARROLL: You know, customers on a node are fed by a CMTS port which has a capacity of 38-1/2 megabits per second in a DOCSIS 2 environment and we measure the traffic capacity of that port and when it reaches a certain point of utilization it's not exhausted, but it starts to impact the performance of our customers. We do not want the performance impacted so we provide relief prior to exhaust. It's as simple as that.
6124 THE CHAIRPERSON: Okay. Thank you.
6125 MR. BHULLAR: So, Mr. Chairman, perhaps I can clarify it for Commissioner Candice Molnar, because she was asking, you know, can you provide us with how you meet once you hit exhaustion.
6126 So Schedule 1 has nothing to do with exhaustion. So that is one of the correlations that was asked for.
6127 THE CHAIRPERSON: We established that.
6128 MR. BHULLAR: Sorry.
6129 THE CHAIRPERSON: Rogers undertook to furnish a new one.
6130 Okay, last question. Over to you.
6131 MR. ROCCA: Peter Rocca. I have a few points.
6132 The first one I wanted to talk about was this whole idea of Elm Street and how, you know, maybe traffic was on Elm Street but not Maple or whatever.
6133 What we are talking about is there is a cost for the town, you know, to use the same analogy, and we may be paying a cost, a unit cost that is divided for that town. We may be paying for usage on Elm Street when we might not even have a single customer on that street.
6134 Likewise, when that street has been upgraded to two lanes, it's not like at 6 o'clock -- if it was upgraded at two lanes, you know, for a peak at noon, why would volume at 6 o'clock in the morning on Elm Street be reflected in costing?
6135 So that's the big issue with kind of that street analogy.
6136 There are a couple of other points, though, regarding Schedule 2. Schedule 2 makes a lot of assumptions under current usage models.
6137 We are trying to -- by using this correlation and saying that, you know, it's volume and not peak we are trying to predict what the future is going to be in terms of usage models.
6138 Likewise, when you have to augment a node, for whatever reason, if it was augmented for upload, download, CPU, it doesn't matter, that capacity has been increased, there is no incremental cost for volume on that node after it has been augmented. I think that's an important thing to keep in mind
6139 THE CHAIRPERSON: But Rogers point that this is the best proxy basically, they are saying we have no exact measurement, but of the various proxies this one most closely resembles actual cost.
6140 MR. ROCCA: Well, there is a lot of problems with the volume model. I mean, what happens in exactly that case when that node has been augmented for whatever reason, does any volume at times that has not been augmented for, you know? So if the peak had to be augmented for noon, now we are being charged for 6:00 a.m. traffic on that link when it's at near zero utilization? It doesn't seem like a very good proxy.
6141 THE CHAIRPERSON: Okay. Thank you very much.
6142 We will take a five-minute break and then we will proceed with Shaw.
6143 Thank you.
--- Recessed at 1220
--- Resumed at 1225
6144 THE SECRETARY: Mr. Chairman, we will now proceed with Shaw Cablesystems G.P. Mr. Jean Brazeau will be reintroducing his panel, for the record, after which you have 12 minutes for your oral rebuttal.
6145 MR. BRAZEAU: Thank you, Mr. Chairman, Members of the Panel.
6146 My name is Jean Brazeau, Senior Vice President of Regulatory Affairs at Shaw Communications. With me today here are my colleagues, Doug McEwen, Vice President of Engineering; and Paul Cowling, Senior Director of Regulatory Affairs. In back is Esther Snow, Manager of Regulatory Affairs
6147 After listening to the presentations during the past week, we are optimistic that a reasonable and balanced solution can emerge from this proceeding that will enhance independent ISP's flexibility without undermining our incentives to invest or distorting facilities based competition.
6148 As we emphasized in our presentation, Shaw's goal is to encourage all our customers, wholesale and retail, to use the Internet to the maximum extent possible. We believe this is the Commission's objective as well.
6149 In order to achieve this, a wholesale billing model must continue to incent investment. At the same time, the model must not distort the intense competition in the western Canadian market. For evidence of that competition, the Commission simply has to look at Shaw's new innovative and customer friendly packages that lead the industry in speed, value and flexibility.
6150 Over the past week, we have carefully considered the relative merits of the primary alternatives being debated, namely the aggregated volume usage model, the peak period model and the MTS Allstream model. As many parties have noted, none of the proposed models is perfect.
6151 As we have emphasized, there are three key criteria for any billing model to work.
6152 First, we must be able to recover our investment through cost-based rates.
6153 Second, the access rate for the wholesale service should vary by speed. As shown in our costing studies, higher speeds cost more because they take up more of the shared capacity of the network components, such as CMTS devices, RF channels, fibre nodes and our hybrid coax fibre infrastructure.
6154 Third, an appropriate line must be drawn between the access and usage portions of the network
6155 Given the shared nature of our network, dividing usage and access is an imperfect science and ultimately comes down to a choice of whether to encourage usage.
6156 Consumers at this hearing have made clear they want to use the Internet more so they can take full advantage of its potential. That echoes what we heard at Shaw's customer consultations.
6157 We continue to believe that the aggregated volume-based model better reflects the shared nature of our cable network. That model recognizes that all TPIA customer usage, not just usage at the ISP's peak, can cause segmentation and other costs.
6158 However, there appears to be support among independent ISPs for a 95th percentile model. If the three key criteria outlined above are met, we are willing to make a 95th percentile model work. There would be implementation costs, as we noted last week, and those costs will affect the monthly TPIA rates.
6159 In response to Commissioner Katz's question, we have determined that our business division, Shaw Business Solutions, does in fact provide transit and transport services which are billed using the 95th percentile.
6160 Shaw's TPIA service uses a separate billing and provisioning system. Therefore, we will still incur IT costs to change TPIA billing to accommodate a 95th percentile model.
6161 In response to Commissioner Morin's request, we have filed an updated comparison of wholesale rates. It is important to note that this is an imperfect comparison because Shaw and MTS Allstream offer different products. It is nonetheless clear that Shaw's rates are very competitive with those of MTS Allstream and the other ILECs.
6162 MTS Allstream has proposed a "capacity-based" model that Primus has also endorsed as its second choice if a peak model cannot be implemented.
6163 The MTS Allstream approach may be suitable for a phone company infrastructure. However, it would be challenging to apply this telco-centric model to a cable network. There is no cable equivalent to an A-HSSPI and we are unable to provide independent ISPs with a dedicated amount of capacity within our network.
6164 We will now speak briefly to certain other issues raised in this proceeding.
6165 With respect to relative penetration levels of TPIA and GAS services, the Commission must be wary of interfering unnecessarily in competitive forces. Based on our knowledge of the market, we believe that if the Commission compares the amount of TPIA users and traffic on our network against the GAS services of all ILECs in western Canada it would find no significant imbalance to redress. The fact is that the Internet market in western Canada is already intensely competitive as is clear from Shaw's new retail packages.
6166 In addition, TPIA rates are competitive with GAS services. This is shown in the comparison we filed in response to Commissioner Morin's request. TPIA speeds are also often much faster and Shaw has always provided speed-matching. Furthermore, Shaw has seen an increase in TPIA customers in the last 12 months.
6167 Clearly, the Commission's determinations in TRP 2010-632 have had a positive effect on the wholesale market.
6168 Regarding POI aggregation, the Commission has already made a final determination in TRP 2010-632. Shaw's TPIA service, POI locations and interim rates are in full compliance with that policy. We provide an aggregated service to all of our TPIA customers.
6169 To be clear, we have not asked, and will not require, any TPIA customer to move its POI.
6170 In conclusion, Mr. Chairman, it is clear that there is no perfect billing model. We continue to believe that the aggregated volume model is most appropriate for our network. However, we are prepared to make the 95th percentile model work as long as the three critical criteria we have identified are met, namely, recovery of our investment costs; rates that vary by speed; and an appropriate demarcation between usage and access components.
6171 Thank you, Mr. Chairman.
6172 We are open for any questions.
6173 THE CHAIRPERSON: Thank you.
6174 You make reference to a response you filed to Commissioner Morin's model.
6175 MR. BRAZEAU: Yes, we have filed that this morning.
6176 THE CHAIRPERSON: Okay. I haven't seen it. I don't have it.
6177 Madam Secretary...?
6178 THE SECRETARY: No, I gave it back to Shaw.
6179 THE CHAIRPERSON: Okay.
6180 You are filing it now then. Okay, fine. So let's file it and make sure --
6181 MR. BRAZEAU: We filed it electronically.
6182 THE CHAIRPERSON: Sorry?
6183 MR. BRAZEAU: We filed it electronically this morning.
6184 THE CHAIRPERSON: Okay. I'm not blaming you --
6185 MR. BRAZEAU: Okay.
6186 THE CHAIRPERSON: -- I just want to make sure it gets distributed, that's all.
6187 While this is being distributed and put on the Web, tell me, you have made now twice the same comment, that you can live with the 95th percentile as long as your three conditions are met.
6188 How do we ensure that? How do we get -- let's assume I take you by your word and say, "Yes, Mr. Brazeau, wonderful, let's do it." What does this mean?
6189 MR. BRAZEAU: Well, I think Mirko raised the concerns that I think all of the sellers of usage-based carriers have is that a lot of the costs that are generated or incremental to the ISP traffic would not be accounted for in our rates that we sell our services to. So that's a significant concern.
6190 But in the costing exercise and in the rates that come out of that costing exercise are truly reflective of our costs, I think we can make any of these models work.
6191 I think the caveat here is that I don't think -- if you do incorporate all those costs, I don't think the formula or the model that you use -- the billing model that we come up with will differ greatly in the rates that we charge the ISPs for their services.
6192 THE CHAIRPERSON: But basically you are not doing much for the ISPs, are you, because effectively you say,: Yes, I will do the 95th percentile, but all the incremental costs I'm going to bill to you. That in the end may not make this a very desirable option for them.
6193 MR. BRAZEAU: Well, again, I don't think we are here --
6194 THE CHAIRPERSON: I mean have you costed out? Do you know what it will cost to do this?
6195 MR. BRAZEAU: The 95th percentile?
6196 THE CHAIRPERSON: Yes. You say:
"If the three key criteria outlined above are met, we are willing to make a 95th percentile model work. There would be implementation costs, as we noted last week, and those costs will affect the monthly TPIA rates."
6197 Obviously. So what are those implementation costs? Have you costed it out?
6198 MR. BRAZEAU: No, we have not. Again, we just assume whatever those costs are would be included in the costing model that we filed with the Commission.
6199 But I don't think those are the more significant costs that have to be accounted for. I think we heard Rogers and the other cablecos talk about all the significant augmentation costs that could occur under any of these models. So those are the real significant costs that have to be taken into account.
6200 THE CHAIRPERSON: Well, but the Cable Companies just said they have no way of -- this is a billing that they wouldn't use for their own customers. So they would have to set it all up, a whole separate billing for TPIA customers and yes, they would have to pay for it.
6201 So without having any idea how much it cost or the timeframe for implementation this is not really a viable option for us at this point in time, right?
6202 MR. BRAZEAU: It would be more costly, absolutely.
6203 We are just trying to find a solution that I think would be amenable to --
6204 THE CHAIRPERSON: You know I appreciate your spirit of cooperation, but all I am saying is we have no idea of what it actually would mean in terms of cost or time.
6205 MR. BRAZEAU: That's correct, and I think you heard from one of the Videotron witnesses that TPIA resale is unique in Canada. And so all of the billing and equipment services that is required in order to provide that service is, again, unique to Canada and therefore when we go to vendors, you know, they look at us and say, "Sure, we can build anything you want but it's going to cost".
6206 THE CHAIRPERSON: Now, I am not sure whether this is a question of language or you mean something. You say AVP is the best model.
6207 Are you referring to the Bell model or you referring to what the Cable Companies put forward just five minutes ago?
6208 MR. BRAZEAU: Yeah, a similar model to the Cable Companies.
6209 THE CHAIRPERSON: Oh, the Cable Companies, that's what you mean, yeah, thank you.
6211 CONSEILLER PENTEFOUNTAS : Monsieur Brazeau, bonjour.
6212 Est-ce qu'on risque d'ouvrir toute une autre boîte de Pandore en changeant d'un modèle basé sur le volume à un modèle basé sur la capacité maximale, parce que ça va déplacer le débat où on va être obligé de trouver et de calculer et de décider sur les coûts de l'implantation de ce modèle-là, à votre avis?
6213 MR. BRAZEAU: I absolutely agree. I think Dennis raised it this morning. If you start billing -- if the Cable Companies start billing like the telcos bill then we are going to have to put a lot of cost at the access level.
6214 And I think Dennis absolutely right. I think when you do that then you are going to create not an opportunity but certainly room for disagreement with the ISPs because all of a sudden they will look at these costs and say, "Wow, I can't pay for these access costs now".
6215 And so I agree. I think if you move -- change the whole billing system that I think this will not be the last ISP hearing that the Commission hears.
6216 COMMISSIONER PENTEFOUNTAS: There was also an international element raised this morning in that the practice, les us et coutumes sont à l'effet qu'on facture en se basant sur le volume et non pas nécessairement sur la capacité maximale.
6217 Est-ce que ça vous préoccupe, le fait que vous voulez être en ligne avec tout ce qui se passe à travers le continent?
6218 MR. BRAZEAU: If that was the case we wouldn't --
6219 COMMISSIONER PENTEFOUNTAS: Is that a legitimate --
6220 MR. BRAZEAU: Well, we wouldn't have TPIA --
6221 COMMISSIONER PENTEFOUNTAS: -- argument?
6222 MR. BRAZEAU: -- because TPIA is unique to Canada.
6223 So you know resale is mandated and I think we proposed the most efficient and effective way to achieve and respond to the requirement for resale. You know, the rates that we have filed are all cost-based rates that have been reviewed by the Commission.
6224 So I think that's -- we can do no more than that.
6225 COMMISSIONER PENTEFOUNTAS: But in your experience the North American custom is to bill on volume and not on peak capacity?
6226 MR. BRAZEAU: For customers, yes.
6227 COMMISSIONER PENTEFOUNTAS: Yes.
6228 Another debate that came up was where we measure, measuring an interface, and does that create a situation? Is that concerning for you that we are not -- if we measure at the interface we are not properly measuring potential peaks along the network?
6229 Mr. Bibic raised that point. Would you like to comment on that?
6230 MR. BRAZEAU: Yes. Doug may want to add some comments on that. He is our engineer.
6231 But I think what we propose or suggest is that you could somehow get a correlation between the peak and what is happening in the network. I'm not sure if the correlation would be as high as the one that was discussed this morning with Rogers but you could find correlation.
6232 But there are risks and, again, I think the model we have today is the most effective model there is given our network and you would take additional risk if you went to a 95th percentile model measuring at the POI.
6234 MR. McEWEN: No further comment.
6235 COMMISSIONER PENTEFOUNTAS: You did such a good job there, Monsieur Brazeau. T'as bien fait ça.
6236 That's fine for me. Thank you.
6237 THE CHAIRPERSON: Len?
6238 COMMISSIONER KATZ: Good morning. I have got two quick questions.
6239 You have identified your three criteria under which you think you can work. The first one says you must be able to recover your investments through cost-based rates.
6240 Do you recover your investments today at the retail level?
6241 MR. BRAZEAU: A couple of things about the retail level.
6242 As we indicated, we have been -- we are certainly facing a significant -- a very competitive marketplace in western Canada. You know, at times as you have seen through our new rates and our new packaging, we are certainly trying to ensure that we retain as many customers as possible. So you will see, you know, some creativity when it comes to rates. Are they all recovering their costs and return on capital? Then on an aggregated basis absolutely they are doing so.
6243 COMMISSIONER KATZ: So the rates that are out there today, the new rates we just launched recently all recover their costs? You are prepared to go to market with those rates?
6244 MR. BRAZEAU: That's right; that's right. I'm just not sure what costs you are talking about.
6245 COMMISSIONER KATZ: Well, the same costs that you are talking about over here "must be able to recover our investments through cost-based rates". So you are recovering it or you are prepared to accept what you are going to market with right now.
6246 Okay, my last question is on page 3. You talk about the Primus model or the MTS model and you say that a similar model doesn't exist on the cable side:
"There is no cable equivalent to AHSSPI and we are unable to provide independent ISP with a dedicated amount of capacity within our network."
6247 I understand that but you are able to provide a certain quality of service no different than you would do to a bank or anybody else as well when you sell them a service. You would provide them with a certain level of guarantee of service levels whether it is five nines or whatever the case may be.
6248 MR. McEWEN: We provide the same level of service to all of our customers. There is no differentiation between customers. Is that your question?
6249 COMMISSIONER KATZ: Well, my question is can you provide Q of S to wholesale customers notwithstanding the fact that you can't provide them with dedicated amounts of capacity similar to the phone companies?
6250 MR. McEWEN: We have refrained from attempting to supply QoS to any customers because of the constraints it would put on the network.
6251 COMMISSIONER KATZ: But you could do it if you had to?
6252 MR. McEWEN: We would be extremely concerned about the impact on the network by attempting to do so. That's exactly why we have never done that, whether it's a business customer or a TPIA customer or a residential customer our service is inherently best effort.
6253 COMMISSIONER KATZ: So you don't provide banks or large corporations with a certain quality of service when you enter into an agreement with them?
6254 MR. McEWEN: Not on a CMTS hybrid fibre-coax network. We only do so when we do it on a fibre-based network.
6255 COMMISSIONER KATZ: Okay. But you do, do it right now?
6256 MR. McEWEN: On a fibre-based network, yes.
6257 COMMISSIONER KATZ: I understand. Thank you.
6258 THE CHAIRPERSON: Candice?
6259 COMMISSIONER MOLNAR: My questions are on that same paragraph, in fact the same sentence:
"There is no cable equivalent to the...[aggregated high speed point of interconnection]..."
6260 And I just -- I need to understand that better because there is a point of interconnection between your network and that of the independent ISPs. So why is that -- just tell me -- explain that to me because I know you said that as well last week.
6261 I asked a couple other parties through the week, could this model be worked with cable? And they said yes. And you have said for the second time, no, it can't work.
6262 So can you explain that to me again?
6263 MR. BRAZEAU: Doug will try to explain it as well as we can.
6264 I think the point we were making here is that you can take all of our network costs and create an access cost if you want to. I mean it's just a cost allocation, and we can do that. But the access costs will be significant.
6265 Right now, the ISPs will bring a GigE that they have leased from somebody else, from a third party into our POI and so we don't charge them for -- there is a one-time like connection charge, and that's it. And then it's based on whatever services or you know they have acquired from us that we bill them on, on a monthly basis.
6266 MR. McEWEN: We are really trying to make, I think, two points.
6267 One is that it is an apples and orange comparison on Commissioner Morin's sheet between a TPIA cableco and a DSL telco because there is a tariff component and one that isn't included in the other.
6268 I think the other point we are trying to make, although there probably is a correlation between the amount of interface capacity that would be ordered under an MTS Allstream model to the amount of network investment we need to do on the access side, well, granted there is a positive correlation coefficient there, it's just not as high as the correlation coefficient between aggregate usage and network operating costs.
6269 COMMISSIONER MOLNAR: I am just going to admit I don't get it. I mean this is a billing -- this is a point for billing, is essentially what it is, right, at the point of interconnection is the point for billing. So can it work for you?
6270 MR. BRAZEAU: As I mentioned, we could make it work. But what you are doing is you are taking the downstream costs and you are putting it back all in access at the point of interconnection. It can be done except that cost --
6271 COMMISSIONER MOLNAR: We are billing it at a point --
6272 MR. BRAZEAU: That cost which is zero today will be whatever the number comes up to.
6273 COMMISSIONER MOLNAR: Okay, very good. And just to go on in that sentence you say:
"We are unable to provide independent ISPs with a dedicated amount of capacity within our network."
6274 Is it your understanding that MTS is proposing a dedicated network?
6275 MR. BRAZEAU: That was our understanding.
6276 COMMISSIONER MOLNAR: I will just maybe ask MTS because my understanding is this is still use of the shared network and it is not a BBM type of an arrangement.
6277 MS GRIFFIN-MUIR: All right. Teresa Muir, MTS Allstream.
6278 That's right. It's part of the shared network. It's not a dedicated network.
6279 COMMISSIONER MOLNAR: Okay, thanks. Those are my questions.
6280 THE CHAIRPERSON: Michel?
6281 CONSEILLER MORIN : Oui. Merci pour vos commentaires, d'abord. Je pense que, grosso modo, c'est des chiffres semblables à ceux que j'avais déposés.
6282 Mais si la Commission devait décider entre le 95e percentile, le modèle de CNOC, et le modèle de MTS, lequel coûterait pour vous le plus cher à mettre en place, même si c'est très challenging, avez-vous dit dans votre paragraphe auquel le vice-président Katz faisait allusion tout à l'heure?
6283 Et deuxième question : De combien de plus? Avez-vous un estimé?
6284 MR. BRAZEAU: I think for the MTS Allstream model the implementation cost would not generate any new billing systems or measuring the 95th percentile cost that we talked about under the 95th percentile. So I think on the billing issue there would be -- the MTS Allstream model would be less costly to implement.
6285 CONSEILLER MORIN : Merci.
6286 Ce sont mes questions, Monsieur le Président.
6287 THE CHAIRPERSON: Anybody in the audience who takes an issue with what Shaw just said?
6288 Okay, Cable Companies.
6289 M. MESSIER : Merci, Monsieur le Président. Je voudrais simplement profiter un peu de l'occasion pour refaire le point sur la question de l'agrégation des points d'interconnexion.
6290 Alors, je note que Shaw n'a pas l'intention d'exiger que ses clients bougent au point d'interconnexion centralisé -- c'est leur choix, et nous le respectons -- mais nous voulons mettre clair que, dans notre cas, nous interprétons la décision comme étant nous n'avons pas cette obligation-là. Comme je l'ai dit ce matin, nous sommes prêts à faire une accommodation très raisonnable pour faire une transition vers les points agrégés.
6291 Ceci étant, nous voulons que... nous vous demandons de lever dans votre décision l'incertitude qu'il y a dans ce dossier-là jusqu'à maintenant à l'effet que certains prétendraient que nous devions maintenir les POI existants. Ceci serait, comme nous avons dit, aller à l'encontre, je pense, de l'objectif de la décision que vous avez faite et créerait une asymétrie supplémentaire avec les ILEC.
6292 Mais cela aussi... nous avons fait le débat l'an dernier. Nous avons plaidé plutôt pour le maintien des POI régionaux. Vous avez plutôt insisté à déployer des POI centralisés. Nous avons mis en place ces POI-là. Nous sommes prêts maintenant à prendre les demandes.
6293 Je pense que de maintenir une exigence ou du moins ne pas clarifier le dossier pourrait perpétuer une incertitude et des disputes où est-ce que nous aurions à supporter des frais supplémentaires. Dans notre cas, il n'est pas question de maintenir deux architectures à ce niveau-là, et nous voulons que... nous vous demandons à mettre ça très clair sur le dossier à la sortie, de façon à faciliter justement une transition vers les POI centralisés.
6294 THE CHAIRPERSON: I think it's clear that there needs to be some clarification of our original decision on that point. What everybody has been asking for, if I understand it correctly, is a transition period and how long, et cetera. We will address that point in our decision.
6295 Anything else anybody else wants? Then we are very behind time but I think it's been a very fruitful discussion.
6296 So let's break now and we will resume at two o'clock. Thank you.
--- Recessed at 1252
--- Resumed at 1403
6297 THE SECRETARY: Ready to start, Mr. Chairman?
6298 We will now proceed with the British Columbia Broadband Association. Mr. Chris Allen is appearing for BCBA.
6299 You may now proceed with your 12-minute presentation.
6300 MR. CHRIS ALLEN: Thanks again, Commissioners, for this opportunity to address the comments expressed by the many esteemed members of our industry. I realize we are fairly tight on time today so I will endeavour to get our organization's main comments out of the way.
6301 We have heard a lot of commentary the previous week on the various methods for accounting for transport over the ILEC and cableco networks. As our members deal exclusively with TELUS at this time I will focus on their comments in this regard.
6302 TELUS currently bundles the transit into the ADSL port price and has stated that they are comfortable with this method continuing. They don't feel there is currently an issue with network capacity and considering that independent ISPs make up approximately 4 percent of the customer base on their network, it would be hard to imagine they would be able to affect their overall capacity in a meaningful way.
6303 The BCBA would be in favour of keeping the existing bundling of transport into the port price provided that TELUS would move to a cost plus mark-up model for pricing that port and transport. This would be simple for TELUS to continue with and wouldn't incur them any new costs which they may in turn have to pass on to their customers. I suspect however that the Commission is looking for a one-size-fits-all model for the industry, although that doesn't necessarily have to be the case.
6304 If we are looking at unbundling the transport component from the port, we believe that transport should be measured using the 95th percentile billing method. We have heard this week from the incumbents that this model will be difficult to implement and from others that it would certainly be expensive and time consuming to evolve to. We do find that to be a bit difficult to understand considering this is very much an industry standard in place and utilized by all the ILECs and Cable Companies for charging for transport on links that they sell for other purposes than aggregated ADSL or TPIA services.
6305 Out of the methods considered, we do consider this method to be a fair way of providing a financial payback to incumbents for the monies spent upgrading their capacity to deliver traffic. By measuring the peak, you are able to capture the cost of that capacity. Models focusing on the amount of gigabytes of actual data moving across the network are unfairly punitive as they compensate an incumbent for off-peak traffic which does not add to the cost of creating that capacity.
6306 The key guiding factor here we continue to believe in is the need for cost-based pricing in this industry. For too long pricing has been based on mark-ups that we believe in many cases are excessive.
6307 In the case of the proposed AVP method Bell has been touting, we believe this mark-up on a per-gigabyte basis to far exceed the actual cost of transporting data across a modern network. We of course are not privy to the actual costing information but one only has to look at incumbent retail pricing to determine that the actual costs to deliver these same services to their retail customers are far, far below what independent ISPs are being asked to pay.
6308 While aggregated ADSL is considered non-essential conditional, we believe it needs to have pricing more in line with essential services. We believe a cost plus 15 percent mark-up on both the naked port and the transport component is a fair mark-up. ILECs will be guaranteed profit and pricing will fall to the point where ISPs can grow their share of the market. At only 4 percent of the market in B.C. and Alberta, we are in a duopoly situation in western Canada for all intents and purposes in the urban markets.
6309 We heard from Distributel last week describing their difficulties with the business model involved in trying to sell TPIA services with their current margins. I think they did well to illustrate this challenge. In the interests of providing an idea of what we believe port pricing needs to be set at to foster competition, I would like to propose some sample pricing that we believe needs to be in place if the Commission is going to see renewed growth in the industry and new entrants.
6310 If the new basic Internet port offering on legacy ADSL networks is to be the 1.5 Megabits to 6 Megabits as suggested in TELUS' tariff applications, it should be no higher than $17 with transport included. If unbundled we would expect to see something like $15.00 for the port and another $2.00 to cover approximately 60 gigabytes of transit or of traffic which is what is provided to a customer typically in a package like that in the retail markets.
6311 As we have all learned from this proceeding neither TELUS nor Shaw charge their retail customers for traffic use currently. For an independent ISP to win customers he will need to be competitive on price and be willing and able to provide substantial included Internet traffic. An ISP would then have to take that port and transport and purchase that same 60 gigabytes of Internet transit out to the Internet itself.
6312 Many Internet providers, for example, often pay $35 per megabit sustained which roughly translates to 320 gigs of usage in a month or $6.56 for that 60 gigabytes of traffic out to the open Internet, or approximately 11 cents per gigabyte. Now, we would need to add the billing, bad debts, technical support and other company costs to the same port which I would conservatively estimate at $5.00 per customer.
6313 We are now at $28.56 delivered for that 1.5 Megabits to 6 Megabits port. An ISP would hope to make $3.00 per month for the customer which would put them at a price point that is very close to incumbent retail in this example.
6314 I think it's worth pointing out that the pricing I'm suggesting here is approximately $10 less than TELUS is proposing in Tariff 391 for the bundled port and that doesn't take into consideration the $100 installation charge as well.
6315 I bring out this above model to illustrate to the Commission that all of the discussions we are having on billing methodologies will be for naught unless the Commission sees fit to provide margins to independent ISPs that will allow them to make sufficient profit to re-invest into their own networks and market effectively to win customers from the incumbents.
6316 I would like to reiterate that whatever model is decided upon for considering the transport portion of Internet delivery that the Commission keep in mind that current pricing structures have not allowed for any appreciable growth in the market share held by independent ISPs. The 4 percent market share in western Canada is especially troubling and will continue to decline without strong action by the Commission.
6317 Any pricing increases, including the consideration of the abolishment of equal pricing for residential and business ports, may completely finish off what is left of this small percentage.
6318 It is our position that, regardless of the decisions made in this proceeding, ultimately the healthiest relationship between ISPs and incumbents will come about about through carrier-specific agreements.
6319 These agreements will allow incumbents and independent ISPs to treat each other truly as business partners looking for mutually beneficial agreements that will allow both parties to prosper.
6320 These off-tariff agreements may well end up forming the future of the industry provided both parties are able to work together and appreciate the important role held in each other's business.
6321 I thank you for your time and look forward to answering any questions you may have.
6322 THE CHAIRPERSON: Thank you.
6323 When you say you believe a "cost plus 15 percent mark-up on both the naked port and the transport component is a fair mark-up" you don't make any differentiation as to what the speed is. I mean we have heard a lot of things that depending on what the speed is there is a different cost and, therefore, the mark-up should be different.
6324 MR. CHRIS ALLEN: Yes. I think -- well, I mean especially on the legacy port side, I think it's pretty clear that there isn't any real differentiation in the naked port piece of it. If you are going to bundle the transit or the transport into the port then I can see the argument made where a port would have a higher cost to the incumbent to provide.
6325 But when you are stripping that transport out of the port the port is really ultimately, you know, on a DSLAM it's simply a port and when you log into that port and set it to the speed, how that actually costs more, I'm not sure.
6326 There may be small amounts of difference but we are not talking about $10 per month per port.
6327 THE CHAIRPERSON: Okay. And you say -- you only repeated part of TELUS' argument which is we should prepare to continue as before but they reserve the right to go to an AVP model and should in effect their wholesale business increase to such an extent that it puts strain on their network or whatever the test is they're applying.
6328 MR. CHRIS ALLEN: Yes. I did notice that in their speech last week as well. They want to have that option, I guess, basically, right? So it does put you guys in a tough position to come up with a --
6329 THE CHAIRPERSON: Your basic position is they don't need that option?
6330 MR. CHRIS ALLEN: I don't -- well, it's simple right now bundled into the port.
6331 Really, I think ultimately the problem is that all this discussion on that transport piece, I really don't believe that it's an expensive piece. Once the capacity is built, you know, it really is just light pulses over a fibre network. It's building that capacity in the first place. Once it's been built and once the investment is paid off, there is no real ongoing cost for that on-net traffic.
6332 THE CHAIRPERSON: Okay.
6333 Marc, you have some questions?
6334 COMMISSIONER PATRONE: Just a few, Mr. Chairman.
6335 Thank you and I appreciate your presentation.
6336 How many of your members are on TPIA right now?
6337 MR. CHRIS ALLEN: Zero.
6338 COMMISSIONER PATRONE: Zero.
6339 Do you have any thoughts about -- at this point having heard all the various arguments about how what is perceived as being an inequity as far as the burden, the wholesale burden on the telcos or on Bell, should be addressed?
6340 MR. CHRIS ALLEN: Like are we more considering TPIA opportunities now?
6341 COMMISSIONER PATRONE: Well, we heard Shaw for instance talk a little bit about the fact that even in the last 12 months that they have seen a growth, an increase in the number of TPIA customers on the wholesale side, but obviously that's none of --
6342 MR. CHRIS ALLEN: It's none of our membership. I would say that I've noticed that Shaw have reached out specifically to my company and have looked -- they seem to be more interested in looking at it.
6343 COMMISSIONER PATRONE: I thought they had a dedicated sales staff knocking on doors.
6344 MR. CHRIS ALLEN: Yes. Well, I didn't hear from them for about six or seven years, so I don't know what was going on at that point in time.
6345 But I think part of it is that a lot of ISPs went from the dial-up era into ADSL in B.C. TELUS, or BC Tel actually at the time, in around '99, started offering the wholesale offerings and there wasn't, to my knowledge, at that time any real opportunity to move into TPIA. And TPIA has been plagued by a lot of technical issues, which I think you guys have tried to address.
6346 If an ISP were starting out today looking at both options, I think that probably TPIA is probably close to as valid anyway as ADSL, but if you have already gone down one path, to then go and duplicate your network by going into the cable path, it's something that ISPs are considering but it's an expensive prospect if you're not looking at getting a lot of customers.
6347 COMMISSIONER PATRONE: It's interesting that you say duplicate because -- is there any scenario under which you might foresee a switch? I can well appreciate that there are costs associated with taking that approach, but do you see any scenario under which that could be attractive for your members?
6348 MR. CHRIS ALLEN: So in terms of switching, like in terms of getting a Shaw connection and then actually changing my ADSL customers over to --
6349 COMMISSIONER PATRONE: Moving your customer base over.
6350 MR. CHRIS ALLEN: It would have to be a compelling price difference, right, but you could certainly envision a business model where if you were at a point where you were breaking even or losing money on your ADSL base and the TPIA offering gave you a path to start producing profit again, then --
6351 Certainly to the end user, I don't think that -- I think there was a comment at one point last week that, you know, do the customers think about themselves as ADSL customers or cable customers. I think they think of themselves as Internet customers.
6352 And so, yes, I could see that scenario, but again, it would come down to the economics.
6353 COMMISSIONER PATRONE: I believe it was you who responded earlier to suggestions that the MTS model leaves service providers vulnerable to gaming.
6354 MR. CHRIS ALLEN: It wasn't me, no.
6355 COMMISSIONER PATRONE: Okay. Do you have any thoughts on that matter?
6356 MR. CHRIS ALLEN: I think the MTS model, you're paying for all the capacity that you would ever use regardless of whether you're using it, which is -- I think it's, in my opinion, probably in some ways even worse than in the AVP model because you could really -- like the models where you buy a redundant pipe, you're paying for a gigabit of sustained traffic on that pipe and maybe you don't even ever use it except for one day when your other feed goes down and you port your traffic out over it for half an afternoon but you could be faced with paying for that the entire year. I think that --
6357 COMMISSIONER PATRONE: That's the CNOC argument, right?
6358 MR. CHRIS ALLEN: Yes. I mean that's -- I don't like that model. Out of all the models, I think that one could be -- in certain situations would be very economically difficult.
6359 COMMISSIONER PATRONE: For the wholesalers?
6360 MR. CHRIS ALLEN: For the wholesaler, for sure. Yes. Yes. If I was -- actually if I was an incumbent, I would see that as a great opportunity to make money because you have an empty pipe that you're charging as if it was full and have it be completely empty for an entire year.
6361 COMMISSIONER PATRONE: What about Bell's argument regarding the fact that there would then be incentive on the part of the wholesalers to increase volume traffic through midnight madness?
6362 MR. CHRIS ALLEN: Yes. I think certainly if you were paying for it, then you would want to try to find some way to use it.
6363 But again, the idea that an independent ISP could somehow control its customer base to get them to do their downloading all throughout the day, I think it's not realistic or practical.
6364 You might get some people by offering free traffic between midnight and six, but in terms of trying to flatten out your Internet usage across the day by controlling your users' usages, it would be very difficult and not practical.
6365 COMMISSIONER PATRONE: I just have one more matter I would like to raise with you, Mr. Allen, regarding your comments about carrier-specific agreements.
6366 MR. CHRIS ALLEN: M'hmm.
6367 COMMISSIONER PATRONE: I believe that this matter was raised earlier, if I'm not mistaken -- and I stand to be corrected by Vice-Chairman Katz -- and the matter was deemed to be delusional as long as small ISPs could always rely on a fallback agreement, a regulated price that might very well be better than anything that they could get through carrier-specific agreements.
6368 Do you want to touch that one?
6369 MR. CHRIS ALLEN: Yes. I think certainly, you know, right now to go into a negotiation with an incumbent to try to produce a carrier-specific agreement while this proceeding is ongoing would be crazy because you may end up talking yourself into signing up for an agreement that's worse in terms of the economics than what the outcome of this proceeding might end up being.
6370 Once this is settled and the pricing is set for probably what will end up being another four- or five-year term or who knows how long, once that's set, then, you know, if it doesn't produce the economics to make you really want to work on your investment, then I think it would encourage you to go to your incumbent partner and say, hey look, you know, we're business partners.
6371 I guess I've always found it a bit ironic that most of the independent ISPs should be considered, I believe, by the incumbents as very valuable customers. The amount of money being sent each month on behalf of their customer base back to the incumbents is a sizeable amount. I think sometimes the adversarial approach that feels like it goes on back and forth, I have a hard time understanding where that comes from.
6372 Ultimately, if I'm Bell and I have an end user who is being served ADSL through my network and is sending their bill back to a Teksavvy, well ultimately, Teksavvy is still compensating me for that port.
6373 Would I rather not have that happen, not have to take care of the billing, the traffic and all the other issues, tech support, all those issues on the port, and just get the money back each month or would I rather have that customer go over to Rogers, for example, right?
6374 COMMISSIONER PATRONE: Is there a result from this proceeding that you think might be conducive to that kind of agreement, that that kind of negotiating might be deemed to be appropriate as a result of what we do here?
6375 MR. CHRIS ALLEN: I think it's difficult for you guys to produce a situation where you kind of foster that agreement.
6376 I think that independent ISPs are more than welcome to those kind of discussions, but I think that everybody has to put their cards on the table a little bit and say, okay, you know, in order for us to be successful, we need to have some margins in our product and what you're giving us is ultimately very similar to retail pricing and you want us to go back and compete against you. It's difficult.
6377 COMMISSIONER PATRONE: Those are my questions. Thank you very much.
6378 MR. CHRIS ALLEN: Okay.
6379 THE CHAIRPERSON: Len?
6380 COMMISSIONER KATZ: Thank you.
6381 Good afternoon. I want to pursue a paragraph you have on here on the last page talking about the Commission seeing fit to provide margins to independent ISPs.
6382 I guess the first question is: Are you a proponent of cost-based rates for wholesale services?
6383 MR. CHRIS ALLEN: Absolutely.
6384 COMMISSIONER KATZ: So how would you see the regime unfolding if we went through a cost-based analysis and after the completion of that the margins that were left for independent ISPs would be too thin, if at all?
6385 MR. CHRIS ALLEN: Well, I think in that case then you would really have to question where the costs were coming from.
6386 I think we're in a good situation that we can look at retail pricing and if a retail pricing for a 6-meg port delivered is $35, then how could you argue that the cost to deliver to a wholesaler would be, you know, substantially similar when a lot of the parts are being stripped out in terms of, you know, the tech support, the billing, the handling, the actual Internet transit itself?
6387 So I think if the numbers being looked at were -- maybe an independent audit of those numbers that were being presented by the incumbent. And I guess in terms of talking about gaming, there's always an opportunity of gaming the cost numbers too when they're being provided.
6388 COMMISSIONER KATZ: A retail-minus model would provide presumably, depending on how much is the minus, would provide a margin for independent ISPs but it may not necessarily be cost-based. What are you views on that?
6389 MR. CHRIS ALLEN: I think the problem with retail minus is -- there's a couple of things really.
6390 Ultimately, retail minus is very hard to administer. Everybody's prices are changing every single month. Bundling, promotions, it really masks the cost of the Internet port. So a lot of things can be done to make it difficult to determine what is the retail that you're minusing.
6391 And then setting that minus rate, I mean certainly 75 percent minus retail, I'm sure we could all make a business model out of that, but I think it would be very laborious for you guys to have to constantly keep an eye on what the retail prices are doing and, you know, if somebody -- there would be billing disputes over what I saw on TELUS' website versus the price I got, what my ports were that month.
6392 I think difficult to administer would be my biggest issue with retail minus.
6393 COMMISSIONER KATZ: So you're confident we can derive a cost-based model that will provide independent ISPs with a reasonable margin while at the same time recovering all the costs that the cable and telcos feel they need in order to justify their business?
6394 MR. CHRIS ALLEN: I am confident that is the case. I believe that you could find a port cost really ultimately, you know, at somewhere around $10 a month, and the transit, you know, $2 or $3. I believe those are real costs and a 15 -- even a 20 percent mark-up on that would still produce a price much less than what the price in the market is today.
6395 COMMISSIONER KATZ: So why haven't any of your members purchased a cable solution?
6396 MR. CHRIS ALLEN: As I said earlier, a lot of them started out on the ADSL path and switching became difficult. There were a lot of technical hurdles at some points, I've understood, in terms of IP management and things of that nature that were better on the ADSL network than the cable network.
6397 That said, like I mentioned, Shaw has been coming to our members with a bit of renewed interest in their wholesale offering side. That may change down the road and especially depending on the decisions here.
6398 COMMISSIONER KATZ: Okay.
6399 You have gone through an example of taking $35-a-month 1-megabit bandwidth, converting it into usage and ultimately getting to 11 cents per gigabyte.
6400 MR. CHRIS ALLEN: M'hmm.
6401 COMMISSIONER KATZ: Can you just take me through it a little slower and explain to me the math that you used to get there?
6402 MR. CHRIS ALLEN: Sure.
6403 A megabit sustained is a measurement of speed but it can be converted to a measurement of actual data as well. By looking at that speed, if it were to be maintained for the entire 30-day month, how much Internet traffic would be used up by that speed?
6404 So the common way that's done is looking at 1-megabit-sustained run for 30 days consecutive on an Internet port would burn 320 gigabytes of data at the end of that month. So from that you can back-figure what 60 would be out of the 320.
6405 COMMISSIONER KATZ: So that is -- when you say "sustained," that is constant usage?
6406 MR. CHRIS ALLEN: Yes. That's if you ran it flat out the whole time. Yes.
6407 COMMISSIONER KATZ: Okay. I understand now. Thank you.
6408 THE CHAIRPERSON: Candice?
6409 COMMISSIONER MOLNAR: Thank you.
6410 I want to talk about the 4 percent in B.C. and Alberta.
6411 MR. CHRIS ALLEN: M'hmm.
6412 COMMISSIONER MOLNAR: We've heard, and I'm sure you've heard, that the West is different and it's highly competitive. And you're operating there, so I suppose you can agree with that?
6413 MR. CHRIS ALLEN: Well, I mean competitive in terms of there's definitely a vigorous competition between Shaw and TELUS in B.C. and Alberta. I absolutely agree with that.
6414 COMMISSIONER MOLNAR: Right. The 4 percent you're speaking of, is that residential, business or a mix?
6415 MR. CHRIS ALLEN: It's a mix and I think that's probably more business than residential.
6416 Residential, particularly in B.C., has been almost impossible for independent ISPs to really penetrate and hold their market shares. Our members have seen nothing but a decline over the last five years, year after year, in our residential ports.
6417 COMMISSIONER MOLNAR: So do you offer residential service?
6418 MR. CHRIS ALLEN: We do both, yes, business and residential. I would say the mix is probably -- at one point when we had our largest amount of ADSL customers, it might have been probably 70-80 percent residential, whereas today I would put that number somewhere around 25 percent of my ADSL accounts at residential.
6419 COMMISSIONER MOLNAR: Okay. So 25 percent residential, and essentially would it continue as a flat rate ADSL service, both the legacy and the FTTN?
6420 MR. CHRIS ALLEN: Would I be in favour of continuing it as a flat-rate service?
6421 COMMISSIONER MOLNAR: For the new speeds as well.
6422 MR. CHRIS ALLEN: Ultimately, I think, yes if the price point -- if the economics were set so that they would work. It's one less thing to worry about if it's all bundled in, and I do believe it is not a large portion of the cost of delivering it.
6423 COMMISSIONER MOLNAR: So what's your value proposition to get the 25 percent res?
6424 MR. CHRIS ALLEN: To get that to increase?
6425 COMMISSIONER MOLNAR: Well, to maintain it, to increase it.
6426 MR. CHRIS ALLEN: I think really ultimately all you can do into that market is win on price. TELUS and Shaw don't charge anything to their end users for Internet usage, so we can't go into the market with an ADSL residential package that has a fixed traffic limit because then it's an inferior product, so you're not going to win on that.
6427 Now, you can win a bit of customers based on your service and, you know, being a local presence in a lot of communities, that counts for something for a lot of people. You can distinguish yourself as being not one of the big guys. You know, that can work for a certain market segment.
6428 But ultimately, you need to be at least at the same price as them and probably a few bucks less to really actually get any penetration, and really, you might have to sell it at cost for a while to get -- at a significantly less price to be able to get them in.
6429 COMMISSIONER MOLNAR: Okay. Well, thank you. I was just trying to understand, you know, as you sit here what your thoughts were related to this residential market in Western Canada and how you saw yourself moving forward in that.
6430 MR. CHRIS ALLEN: Yes. It's very difficult. I think, you know, if --
6431 COMMISSIONER MOLNAR: Is that where you're looking or are you looking more to the business?
6432 MR. CHRIS ALLEN: To be honest, most of the companies in our organization have changed their focus over the last few years towards rural wireless deployments.
6433 That's where all -- almost all of our investment dollars have gone, and my company specifically has spent millions of dollars in infrastructure, into building towers and spreading out rural Internet into areas around the province that can't get wireline services.
6434 And that's a good market because you can put your money in, you know your costs, and you don't send any money back to the ILEC except for your traffic that you're still buying off the ILEC. But other than that, you know, each wireless customer sends me back their monthly amount and I keep that amount instead of the model in ADSL where literally 80-90 percent of what the customer sends me goes right back into the ILEC pocket.
6435 COMMISSIONER MOLNAR: Retail is not part of this proceeding, as you well know --
6436 MR. CHRIS ALLEN: No. I understand.
6437 COMMISSIONER MOLNAR: -- but I am just interested whether or not you have flat rates on your rural wireless.
6438 MR. CHRIS ALLEN: No, we do not for a couple of reasons.
6439 One, the Internet traffic to deliver it out into rural areas is more expensive than it is in urban areas.
6440 Two, the wireless network infrastructure can't handle people running 60-70 gigs of traffic a month over it without causing congestion a lot quicker than a wireline network would.
6441 So we have fixed caps and traffic costs. So we do UBB billing into our rural markets.
6442 COMMISSIONER MOLNAR: Okay, thank you. Those are my questions.
6443 THE CHAIRPERSON: Okay, thank you very much. Those are all our questions.
6444 MR. CHRIS ALLEN: Okay. Great. Thanks, guys.
6445 LE PRÉSIDENT : Madame la Secrétaire, let's proceed with the next one without a break.
6446 THE SECRETARY: I would now invite TELUS to take place.
6447 THE CHAIRPERSON: Monsieur Perrier?
6448 THE SECRETARY: We will now hear TELUS Communications Company, Mr. Ted Woodhead.
6449 Please reintroduce yourselves for the record, after which you have 12 minutes for your oral rebuttal.
6450 MR. WOODHEAD: Thank you, Madam Secretary.
6451 Good afternoon, Mr. Chairman, Vice-Chairs, Commissioners. I am Ted Woodhead, Vice-President, Telecom Policy and Regulatory Affairs.
6452 With me on the panel:
6453 - to my immediate right, is Zouheir Mansourati, Vice-President, Network Technology and Planning;
6454 - to the right of Zouheir is Orest Romaniuk, Vice-President and Controller, Finance; and
6455 - to my left is Eric Edora, Director, Regulatory Affairs.
6456 We thank you for the opportunity to make this presentation.
6457 TELUS has listened to the concerns and issues raised by all of the participants in the hearing since last week. It has paid special attention to the concerns raised by the many consumers that have spent time and money to make their views known to the Commission. They have shown just how fundamental the Internet is to their lives and we know that they represent a cross-section of consumers across Canada.
6458 TELUS believes that its Internet services, both in the retail and wholesale marketplaces, are consistent with what Canadians want. You have heard that we offer retail services with substantially higher data packages than most North American carriers, topping out at 500 GB per month and available on a standalone basis.
6459 On the wholesale side, our proposals are for flat-rate plans at all speeds, both legacy and speed-matching. Therefore, we would not force our wholesale ISP customers to charge usage-based billing to their end customers because they would pay the monthly wholesale charge with no fixed amount of volume attached. We hope that the CRTC continues to allow network providers the flexibility to offer flat-rate wholesale plans.
6460 Based on our proposed tariff, wholesale ISPs would be free to set their packages to their residential customers in any way they wish. That includes the ability to offer unlimited plans. However, if they do, we should be able to charge them additional costs should it turn out that their unlimited or other extremely high-usage retail plans cause additional costs on our network, other than the costs built into our tariffs. Therefore, we do not oppose the principle of applying volume charges and reserve the right to adopt the model for usage-based billing that the CRTC approves.
6461 The Commission has asked which of the models proposed should be approved for volume-sensitive rates. TELUS does not generally oppose any particular model.
6462 Having said that, TELUS notes that two fundamental principles should guide the Commission's determinations. First, the model must be relatively simple to implement, easy to understand and transparent to the ISP customer. Second, the model must allow for full recovery of costs and a suitable mark-up.
6463 For TELUS, an aggregated-volume type of model is more consistent with these principles than the 95th percentile model. It is easier for the Company to implement, is simpler to administer and easier to communicate with customers. Moreover, this aggregated-volume model appears to be the one that would potentially allow the Company to better recover the costs associated with network congestion that would occur on TELUS' shared portion of the network.
6464 Having said that, TELUS does not support a usage charge that is levied in advance with no recourse or credit for unused capacity. In order for an aggregated-volume-pricing model to be transparent to the ISP, any volume charge should be levied in arrears so the ISP only pays for data that its customers have consumed.
6466 MR. ROMANIUK: Thank you, Ted.
6467 On the issue of the mark-up, the Commission must recognize that the residential Internet service market is fundamentally different versus traditional monopoly telephone service. Virtually all of the industry parties recognize that there is significant competition for these services and accordingly inherent risk in these investments.
6468 In TELUS' case, our investments have cost hundreds of millions of dollars and this investment will continue.
6469 Broadband investments must be made by private companies, and private companies will only make these investments if there is an opportunity to recover costs and make adequate returns.
6470 That is why the mark-up on residential wholesale Internet services is so critical. It doesn't represent pure profit margin. It represents:
6471 first, a proportionate recovery for fixed and common costs;
6472 second, compensation for the risk of investments made in the network facilities to provide these services, and
6473 finally, an adequate return to shareholders.
6474 All three of these components are critical to maintain current and future network investments. TELUS' proposed mark-ups would recover all of these components.
6475 CNOC has proposed that any mark-up be assessed at 15 percent over Phase II costs. This is the defined mark-up that is reserved for essential and conditional essential facilities.
6476 The Commission has consistently ruled that ILECs' aggregated ADSL services are not conditional essential. In the most recent review of these services, the proceeding that led to Decision 2010-632, the Commission did not change the categorization of these facilities as conditional mandated non-essential. In fact, in that decision, the Commission recognized an additional 10 percent mark-up was required for the ILECs' matching speeds.
6477 Therefore, the Commission cannot apply CNOC's proposed mark-up without reviewing and varying both the Essential Services and the Matching Speeds decisions.
6478 TELUS would like to say some final words about the costs of investment.
6479 We have provided data that shows that data usage by retail end-customers will only increase in the future and wholesale end-customers will be no different. No matter what any party tries to tell you, 40 percent annual growth or more is huge and it shows no signs of abating. It costs money to respond to this continual increase in demand. TELUS anticipates the need to continue to invest in network build to address this usage growth.
6480 Furthermore, parties that tell you that costs for network equipment are falling miss a big part of the picture. In our opening statement, we noted that we are deploying DSLAMs deeper in the network and considerably closer to the end-customer.
6481 Engineering, furnishing and installation share of the overall cost is much higher then when DSLAMs were being installed in central offices. Our analysis shows that while the cost of a DSLAM has decreased, the associated installation, space and power and network connectivity costs have increased such that the overall costs are significantly higher on a per-end-user basis. When you add in the factor that we are installing more and more DSLAMs, you can see how our DSLAM costs have increased tremendously to address the increased capacity requirements.
6483 MR. WOODHEAD: We understand the competing views that have been offered throughout this proceeding. TELUS goes back to what should be the overriding objectives that caused this proceeding: Canadians want access to the Internet with the freedom to use it as they wish and they want their choice of provider.
6484 The best way to achieve these outcomes is to ensure that network providers continue to build and enhance their networks, to deliver the highest speeds to customers.
6485 From a wholesale perspective, TELUS' proposed tariff maintains the incentives to invest and allows wholesale ISPs the ability to provide their own high-speed Internet services using TELUS' network and to create any retail Internet packages that they wish.
6486 That concludes our rebuttal presentation. We will be pleased to answer any questions you have.
6487 THE CHAIRPERSON: Thank you.
6488 Explain paragraph 4 to me. I'm not quite sure I understand it.
6489 You say you want a flat rate plan and then you say:
"That includes the ability to offer unlimited plans. However, if they do, we should be able to charge them additional costs should it turn out that their unlimited or other extremely high-usage retail plans cause additional costs on our network..."
6490 Operationalize that for me. If we accept this idea, what do you expect us to do?
6491 MR. WOODHEAD: Well, how I would see this operating is today we don't see that we are having a congestion issue, therefore we are satisfied with the existing tariffs that we have proposed.
6492 In the event that that usage, that 40 percent usage that we are seeing across the profile, including the wholesale piece, if it were to get to a point where it was causing us congestion at various parts in the network, then of course we would come back and look to you, presuming that there is something that flows from this proceeding and ask for approval of some plan consistent with rules that you set out.
6493 THE CHAIRPERSON: So you would in fact come to us and ask to review the tariff tat we set for you?
6494 MR. WOODHEAD: If that's what was required, yes.
6495 THE CHAIRPERSON: I see, okay.
6496 I think you are the only one, at least to my knowledge, that has suggested that the payment would be after rather than before the usage so that you can take into account under use and give the people credit, et cetera.
6497 Was this in your original proposal?
6498 MR. WOODHEAD: No, it wasn't. This is new, I mean in the sense that it was responsive to -- I think what you heard from some parties was that in the prepayment plan they would only perhaps be using some smaller amount and that they felt that that was unfair and so we were trying to respond to that by saying pay post and pay for what you use.
6499 THE CHAIRPERSON: Is this in answer to Mr. Bibic's fear that if we adopted the MTS model it would in effect give everybody an incentive to use as much as possible, because after all they have to pay for it, they might as well get their users to use it.
6500 If indeed you are only paying after the fact and you get a credit for what you actually used, would that not answer the fear that an MTS model would encourage and incite wholesalers to maximize their usage?
6501 MR. WOODHEAD: It may in fact be a response to that, yes -- an answer to that, sorry.
6502 THE CHAIRPERSON: Thank you.
6503 Tim, you had some questions?
6504 COMMISSIONER DENTON: Mr. Chairman, I have very few since you proposal seems to avoid most of the problems.
6505 It may be superfluous to ask this, but I notice that you say in paragraph 6:
"... aggregated volume model appears to be the one that would potentially allow the Company to better recover the costs associated with network congestion ..."
6506 And so forth.
6507 This is the only point on which you seem to be making a statement different from several other players in the room. Would you care to elaborate on this?
6508 MR. WOODHEAD: I will take a brief crack and then I will pass it over to --
6509 COMMISSIONER DENTON: Pass it to the expert.
6510 MR. WOODHEAD: -- to Zouheir.
6511 The reason we would say that is because we do believe that there are various points in the network where congestion occurs -- and I think you have heard Mr. Engelhart speak to it, you have heard Mr. Bibic speak to it and some others -- which are not, in our view, picked up by the peak pricing model.
6512 So, Commissioner Denton, that's the genesis of that statement, but I will let Zouheir augment that.
6513 MR. MANSOURATI: Sure. I would like to bring to your attention, Mr. Commissioner, that we don't use the term "congestion" in our proposal and the reason is simple -- excuse me, my throat is a little bit hoarse.
6514 The reason is simple. We don't believe that billing mechanisms should be used to do network planning and traffic engineering. These are two separate things in our mind and we believe that proper network planning and proper traffic engineering would be separate as an activity that we follow.
6515 Therefore, we don't use billing as a means to do congestion control.
6516 COMMISSIONER DENTON: You are perfectly lucid in stating that, but why should billing mechanisms not assist you in network planning?
6517 MR. MANSOURATI: We are simply stating that in an imperfect world, and two imperfect solutions, we have a preference and that preference is towards the volume because we see it as simpler and still achieving the same result.
6518 We are not saying that the other one will not achieve a similar result.
6519 COMMISSIONER DENTON: Okay.
6520 Mr. Chairman, I don't have further questions for these people.
6521 THE CHAIRPERSON: Okay.
6522 Mr. Bibic, can I ask you to comment on what Mr. Woodhead has just said, that the ex-post billing rather than advanced might be an answer to your fear that an MTS model or anything akin would incite over use?
6523 MR. BIBIC: Well, Mr. Chairman, you will hear in a few minutes when we are up that we mentioned this morning we agreed to cost-based AVP rates for both legacy and FTTN. That was one point I made this morning.
6524 We also believe we now will come forward with a proposal for one AVP rate, thus eliminating the need for the prepaid and post-paid structure.
6525 So we are okay with -- we are in alignment with TELUS that AVP we would bill in arrears, so whatever is used we will bill for at one AVP rate, not one rate for what you guess in advance and a higher rate for what you have actually used afterwards.
6526 THE CHAIRPERSON: I see. Okay. Thank you. I will look forward to your testimony.
6527 Does anybody else have any questions for TELUS?
6528 COMMISSIONER PATRONE: I have one.
6529 THE CHAIRPERSON: Go ahead.
6530 COMMISSIONER PATRONE: Just one question, Mr. Chairman, and it has to do with paragraph 10 regarding where you talk a little bit about investment and you say:
"No matter what any party tries to tell you, 40% annual growth or more is huge..."
6531 Then you go on to say:
"It costs money to respond to this continual increase in demand."
6532 Is there a regime that would perhaps either dampen or incent rolling out upgrades to your network going forward that we should take with us as we go forward with it?
6533 MR. WOODHEAD: I will let both Orest and Zouheir answer and then I have a view, which I may or may not share.
6534 MR. MANSOURATI: Again, I would like to just restate the fact that we don't believe billing is going to drive the capacity. We live in an ILEC territory in any case, it is very competitive in nature and for that reason there are two dimensions that need to be taken into account, the competitive dimension, but also the growth in demand.
6535 On the growth and demand perspective it's pushing us to get closer and closer to the end user, so we went from CO to FTTN, possibly having to go deeper with DSL. We are also introducing solutions that will require us to go back far earlier than we saw in improving and upgrading the network to take advantage of things like bonding of two copper pairs or introducing vectoring as a solution to improve reliability and to improve the rate.
6536 That could also push us to deploying fibre to the home or fibre to the prem, Under G-POD for example. These are investments that are long in their return periods and therefore we are planning, according to these growth rates, 35 to 40 percent, but when you look at it from a competitive perspective it's difficult for us to be able to apply rates that would shorten that return on investment period.
6537 I'm not sure if that addresses your question.
6538 MR. ROMANIUK: Just also I would like to add, when we are looking at the 40 percent growth rate, say the average life today of a DSLAM is 6 years, 40 percent per year in year 6 results in 752 percent growth. That is the magnitude. So 752 percent higher than what it is today. Your network cannot handle that today so it will require new investment probably before year 6. That's what we are saying.
6539 MR. MANSOURATI: Yes.
6540 COMMISSIONER PATRONE: So it would shorten the duration --
6541 MR. MANSOURATI: The life cycle of equipment, yes.
6542 COMMISSIONER PATRONE: -- of the lifespan of the DSLAM.
6543 MR. MANSOURATI: Yes.
6544 COMMISSIONER PATRONE: Did you decide, Mr. Woodhead, whether you wanted to share or keep it to yourself?
6545 MR. WOODHEAD: No, that hit it pretty nicely.
6546 COMMISSIONER PATRONE: Thank you.
6547 THE CHAIRPERSON: Okay. Thank you very much. Those are out questions.
6548 We will take a -- sorry, there is somebody up there.
6549 MR. TACIT: Yes. I just wanted to address --
6550 THE CHAIRPERSON: Identify yourself, please.
6551 MR. TACIT: Oh, sorry. It's Chris Tacit, CNOC.
6552 I just wanted to address two points.
6553 First of all, on the mark-up issue, we have not suggested 15 percent overall, it was only on the access portions, and that was on the basis of how unbundled access has been treated by the Commission in the essential services decision.
6554 We view that any rate restructuring that might occur in this proceeding would certainly allow the Commission to re-look at the kinds of mark-ups that it should apply to both access and transport because it would be a new rate structure, it wouldn't necessarily be the same one that we have had before.
6555 That's on that point.
6556 Second, on the flat rate issue, you know I have heard the Commission asking a lot of questions about what it would take to incent greater use of TPIA, but there is another big problem and that is that across the country the only ILEC in whose territory there is some degree of competition really that's measurable, non-trivial, is Bell Canada.
6557 TELUS, MTS Allstream, Bell, Aliant Atlantic, SaskTel have very, very negligible amounts of competition from other than the cable company and we believe the reason for that is the existing flat rate structure. That's why we have been advocating an unbundling of transport from access.
6558 THE CHAIRPERSON: Okay. Thank you.
6559 Back there in the last row?
6560 MR. BHULLAR: Hi. This is Jatinder Bhullar.
6561 I just have a question for TELUS on paragraph 3 about 500 gigabyte usage being highest in North America.
6562 So if it is coming from a person like a residential customer paying $50, that translates into all-in $.10 a gigabyte.
6563 Does that help you make profit? I assume it does, so isn't that a way to also look downstream, that this has gone all the way to the Internet where you are paying transit fees and everything else, which you don't incur for wholesale, so is $.10 the upper limit on what should be charged to a wholesale?
6564 MR. WOODHEAD: I will let Zouheir take a crack at that.
6565 MR. MANSOURATI: Yes. I mean let's be careful not to extrapolate too far. We are talking about the extreme cases, the high users, translating that into price per byte would be not constructive in the least.
6566 This is just talking about the extreme case, so you can't derive from that cost per gigabyte or per byte.
6567 MR. BHULLAR: Then again like the issue is, I can get 20 or 50 to make one gigabyte five as a residential customer and then create a wholesale business model to get equivalent capacity. So that's where I was flagging it.
6568 Thank you.
6569 THE CHAIRPERSON: Okay.
6571 MR. BIBIC: Yes, thank you, Mr. Chairman. Bibic, for Bell.
6572 I just want to respond very quickly to Mr. Tacit's point where I think he just said that CNOC advocates a mark-up of 15 percent for access.
6573 Of course that was precisely what was decided last year in the speed matching decision, that access is not an essential service. So while a mark-up isn't scope for access in this proceeding, a mark-up of 15 percent is not in scope. That was decided last year.
6574 This service is conditional mandated non-essential wholesale service and that was in paragraph 57 of last year's decision.
6575 THE CHAIRPERSON: I believe TELUS made the same point in their presentation.
6576 MR. ROMANIUK: Yes, and just on that point, 15 percent does not even come close to recovering fixed and common, never mind the risk. If I would be looking at, say, a risk pendulum this might be a little bit more on the riskier side versus something else, so 15 percent on your fixed and common is not there, never mind a return to shareholder.
6577 THE CHAIRPERSON: Okay. Let's take a 10-minute break and then we will start with Bell.
--- Recessed at 1453
--- Resumed at 1506
6578 THE CHAIRPERSON: Okay, commençons.
6579 THE SECRETARY: We will now hear the oral rebuttal of Bell Aliant Regional Communications Limited Partnerships and Bell Canada, collectively The Companies.
6580 Appearing for The Companies is Mr. Bibic.
6581 Please reintroduce yourselves for the record, after which you have 12 minutes for your oral rebuttal.
6582 MR. BIBIC: Thank you, Madam Secretary. Good afternoon.
6583 I have with me, to my immediate right, Jonathan Daniels; to his right Carl Condon; to Carl's right James Gilmore, James is also with Network Technology and Planning; and to my left, Phil Gauvin, and in the back, well, a whole bunch of people. I think I will just move forward.
6584 MR. BIBIC: So I'm at paragraph 2.
6585 We have asked for the regulatory authority to implement an economic ITMP, and of course economic ITMPs are authorized and they are indeed the preferred CRTC ITMP.
6586 Just because users, whether retail or wholesale, do not typically experience congestion does not mean it does not exist, it means that we have successfully resolved congestion problems before they become noticeable to users. That's a point of pride for us.
6587 We do this by making massive investments every year to relieve congestion. As we mentioned last week, our capital intensity ratio is greater than 100 percent on the Internet. This means that our Internet capital expenditures exceed our Internet revenues.
6588 But we have to monetize these investments if we are to have the incentive to continue our network rollouts. So the key question is this: Should the economic ITMP be based on peak or capacity pricing at a single point of interconnection, or volume pricing to better reflect usage impacts across a shared network? Or course everyone knows that we favour the volume pricing model.
6589 Now, first off, before I continue, I want to say that we do agree with cable's submissions regarding the impacts on the network that would be caused by 95th percentile. We do think that 95th measures the wrong thing at the wrong time.
6590 Completely agree with Ken Engelhart's analogy with Maple Street and Main Street, et cetera. I think he said quite succinctly what I was trying to say several times this morning.
6591 So just for the structure of this, the text I have here is focused specifically on the points I was trying to make this morning in terms of cost recovery. Once I'm finished the text I'm going to move to the diagrams that are also attached to the opening statement and that focuses more on what cable was also explaining this morning on how 95th or capacity pricing measures are the wrong thing at the wrong time. So that is how we have structured this.
6592 So now paragraph 5.
6593 Wholesale ISPs say they are prepared to pay for usage. CNOC wants 95th percentile. Primus suggests total capacity pricing, agreeing to pay for even more than they actually use. Last week this prompted Vice Chair Pentefountas to ask Primus:
"You made a compelling case for peak traffic billing, and if I were Bell, I would go with peak traffic billing, if we were to assume that what you say is correct. Why wouldn't they?"
6594 Indeed, Commissioners, where is the catch? Why is it that wholesale ISPs suddenly are prepared to pay for all of their peak usage or, according to Primus even pay for full capacity, regardless of what they actually use?
6595 The simple answer: They won't! Their model is premised on two principles completely at odds with each other:
6596 One, if network providers, like us, charge to the peak, the wholesale ISPs say it doesn't matter if they drive up total usage on our shared network, since they are already paying for the whole month at the peak price; but
6597 At the same time, the wholesale ISPs demand that wholesale rates should be sufficiently low to guarantee their viability, regardless of the extra cost the ISPs' overall total usage imposes on the incumbents.
6598 I think that was pretty clear last week and even today
6599 With 95th percentile or capacity pricing, an ISP will have the incentive to use at or close to its peak at other times of the day. In other words, to flatten their traffic distribution to the highest level possible.
6600 What we mean here is actually not reducing their peak, but actually adding traffic at other times so long as it doesn't hit the peak or exceed the peak or exceed the capacity, depending on which model you are talking about.
6601 Now, moving traffic -- not moving traffic, but adding traffic to other times of day. You gave an example of the midnight madness sale and there are several questions coming up: Is that really feasible?
6602 Well, between when I said it and over the lunch break we did some research and in the U.K. there are several ISPs who do this. For example, Slingshot has capped, but if you use the Internet between 1:00 a.m. and 7:00 a.m. that usage doesn't count against your cap. Eclipse does the very same thing, different time period, 11:00 p.m. to 9:00 a.m. and another ISP called Plusnet also does the same thing.
6603 So it could be at midnight, it could be at any time, it could be 4:00 p.m., 3:00 p.m., none of that matters. It could be due to cloud computing or whatever the consumer wants to do. You can call it innovation as Mr. Stein calls it, it doesn't matter to me, the point is that usage should be paid for.
6604 So now I'm at paragraph 8, Commissioners. I'm kind of jumping around a little bit because some of what I have in the text has already been covered.
6605 The incentive for ISPs to drive more traffic or flatten their traffic distribution to the highest level possible will cause us to incur significant costs to relieve congestion on multiple links in the aggregation network. We tried to show that last week with our Figures 1 and 2, and we have another figure that tries to do the same thing, and I will cover that a little bit later.
6606 The problem is that Bell will not be compensated for the increased costs caused by an ISP causing higher overall usage.
6607 When I say higher costs, it's not just implementation costs. In fact, I suspect that the implementation costs are just a small -- they are big, but a small portion of what we are really talking about in terms of the costs. We are talking about the costs of augmenting multiple links in the aggregation network because of this.
6608 By the way, I would also add that this notion of the ISPs driving more traffic, since they have already paid for it, or they are not going to pay more, we have been calling this gaming, but I don't think that gaming is the right label for it. It's not really gaming, it's about an ISP trying to make the most efficient use of what they have already paid for, whether it's 95th or capacity pricing.
6609 So there is that label there that is a little bit pejorative, which I think we should stop using.
6610 My point was, we won't be compensated for these costs, and this is because regulatory costing is based on how we build our networks. We do not build our networks as if they are in use all the time at peak or at capacity. Last week, Mr. Cohen acknowledged that his end users could cause congestion at times other than his overall 95th percentile peak, and that that's not reflected in the rate that ISPs pay Bell today.
6611 Now, I covered that this morning. I will skip over the quote, but it's right there, where Mr. Cohen says, in response to Mr. Pentefountas, that, yes, his user could be causing peaks at other times of the day, and that is not reflected in our rates.
6612 Paragraph 10: The ISPs assert that the easy solution is to adopt cost-based rates and to reflect the extra costs in the wholesale rate, but only if the wholesale rate remains sufficiently below the incumbent's retail price.
6613 In Mr. Tacit's words last week, you should focus "on the actual dollar amount at the end of the day, because that's what really matters."
6614 We heard something, actually, quite similar just an hour or so ago from the B.C. Broadband Association, which pleaded for margin protection.
6615 The catch is this. One, with 95th percentile or capacity pricing, the wholesale ISPs will have the incentive and ability to drive more usage on our shared network. This is clear from Mr. Stein's testimony.
6616 By the way, I also have a chart to show the incentive and ability. I will get there in a second.
6617 Two, this will cause more costs for us in the aggregation network, costs not reflected in the rates that wholesale ISPs pay us today.
6618 Now, when our customers cause the congestion at Points 1, 2 and 3 in our diagrams, we bill them. When wholesale end users cause that congestion, today we can't.
6619 Three, while the wholesale ISPs say they are prepared to pay a cost-based wholesale rate, which, in theory, could allow us to recover extra costs, they are comfortable in the knowledge that the Commission will set wholesale rates with a sufficient margin to ensure their viability, and that risks providing yet another opportunity for regulatory arbitrage.
6620 So, incentive to use more, extra costs in the aggregation network, inability to recover, and that is the catch.
6621 In comparing the models thus far, the Commission has assumed that they are revenue neutral to Bell. We can make that assumption, but they are certainly not cost neutral.
6622 Again, that's the catch.
6623 There is absolutely nothing wrong with more usage of the Internet, but those who use more should pay more for their usage. AVP ensures that this will occur, since it attracts usage rates directly to actual usage, nothing more and nothing less.
6624 Again, to borrow from this morning's analogy, it will count every single car, not just the cars on Main Street at 8 p.m., which is an imperfect proxy.
6625 Everyone understands it. Consumers understand it across many industries. We just heard the BCBA say that when it is networks that they own, which are wireless networks, and there is congestion, they charge usage-based billing at retail. That's exactly the point. We have to find the right model for wholesale now.
6626 To conclude on the written text, we are prepared to make the following significant modifications to further simplify our AVP proposal. We agree to cost-based AVP rates for legacy and FTTN, with the same mark-ups as for access; again, but not 15 percent.
6627 This results in one AVP rate of 17.8 cents per gig per month. So that eliminates the pre and post-paid rate distinction.
6628 And we will bill all ISPs in arrears based on actual traffic models.
6629 Before I leave this and go to the diagrams, Mr. Chairman, I just want to reiterate that, under our model, AVP also applies to legacy, and in that respect it is very, very similar to the cable model. Of course, there are some details that differ in implementation and the actual rates, but the Bell legacy AVP model and the cable model are practically the same.
6630 I was hoping to turn to the attachments, and Figures 1, 2 and 3 are the same figures that we provided last week, so I won't go through those. They are here just to make sure we have a complete set.
6631 The new ones are Figures 4 and 5.
6632 I suspect that I am going to run out of time here, so I could end it here and we could have a dialogue, or I could just keep going.
6633 THE CHAIRPERSON: This is all about Bell, this hearing, so you can have as much time -- because, after all, you are the biggest supplier to ISPs of wholesale traffic.
6634 So I am not going to cut you off.
6635 MR. BIBIC: Thank you, I appreciate that, and, actually, this is very germane to the discussions we have had today.
6636 If we start at Figure 4, at the top half of the page, call this the default. This is simplistic, of course, but I am just trying to emphasize a point.
6637 And let's assume that you know this diagram, you saw it last week.
6638 The top starts at 8 p.m., and I am assuming that 8 p.m. is the ISP's peak under 95th percentile.
6639 If you are at the far right with me, you will see that there is the AHSSPI, the interface. We have assumed here that the ISP has purchased a 1 gig pipe, and at peak its 95th percentile measurement is 600 megabytes per second. So it is well below the capacity of the pipe, but that happens to be its peak.
6640 How is that peak generated at 8 p.m.? Move with me all the way to the far left. It's very simplistic. We have four COs serving eight homes here. The link off the top CO, CO A, at its peak the throughput is 250 megabytes per second. You will see that it is lower for the second home, at 100, 50 for the third CO, and 200 megabytes per second for the fourth CO.
6641 If you add them all up, that is how you get your peak throughput of 600.
6642 So that's the baseline.
6643 The point here is, again on the far left, we have provisioned these links -- the maximum link -- take the link at the very top, the one that says 250. We have provisioned that link to operate at 250, and we have costed it that way.
6644 So that would be the baseline.
6645 Now, what I am trying to show here, having set that as the template -- go down to the bottom half with me. Here we are looking at 1 a.m. I am just trying to show the whole point that we have discussed time and again about -- use my midnight madness example.
6646 At 1 a.m., assume that this fictional ISP has implemented one of these midnight madness sales, à la U.K., and that actually causes, again on the far left here -- you see there is a link that is now red. The midnight madness sale encourages the homes served by that Central Office to actually use more. So now they are using 400 megabytes per second at 1 a.m., and we have provisioned for 250. It's now 400. We have to go there and augment that link.
6647 But if you add up -- if the second Central Office is functioning at zero, that's because people are asleep at 1 a.m. The next Central Office is functioning at zero. They are asleep. Hypothetically, in this example, the fourth CO -- that link is operating at 100. They are interested in midnight madness, or they are busy, they are working late.
6648 If you add them all up, it's 500, well below the 600 megabyte peak, but still they have caused a jump in one of the links that we have had to augment.
6649 That is the point that we were trying to make this morning and last week, and that I think cable was also trying to show.
6650 That is the first thing that I wanted to explain with this chart. Now I want to kind of map the Rogers' correlation with a correlation that I have in Figure 5, but stay with me on Figure 4 for a second.
6651 The links that I have on the far left, which run from the Central Offices to the Metro Network, what Rogers showed, I think -- well, I know -- is that total volume is a good proxy for the peak that is occurring at their nodes or our DSLAMs.
6652 What they have shown is that high volume correlates with peaks on these nodes. So that's what their Schedule 2 showed.
6653 As I mentioned when I interjected, what that establishes is that peaks are caused by traffic volumes at those nodes, or at our DSLAMs. So now the question is: Do the peaks at the AHSSPI, on the far right, correlate with volume as well in the aggregation network.
6654 The answer is no.
6655 So what we are establishing in Figure 5 is that there is no correlation between the peaks at the AHSSPI and what is happening in the aggregation network.
6656 So if we turn to Figure 5, you will see here that there is no relationship at AHSSPI between 95th peak throughput and total usage. A high 95th percentile -- go, if you will, on the Y axis, to between 12,000 and 14,000 megabytes per second, which is at the high end, and walk toward the right with me across the chart. You will see that -- call it a 13,000 megabyte per second 95th percentile can result from about 40,000 gigs of total daily usage, if you go down to the X axis. It could equally cause 10,000 gigabytes of total daily usage, as well.
6657 So the same peak can cause 40,000 gigs of total daily usage or 100,000, more than double.
6658 Conversely, if you go down to the bottom right, you will see that a low 95th percentile of around 4,000, or slightly below 4,000 megabytes per second, can result in less than 20,000 total gigs of daily use, or more like 50.
6659 So there is no correlation, then, between the peaks at the AHSSPI and total volume, and hence no correlation to what is happening in the aggregation network.
6660 I would point out here that what we did in order to map these points out is, we took six ISPs who are customers of ours -- this is actual data -- for every single day in the month of June 2011 and we correlated their peaks in a day versus their total usage in a particular day.
6661 That's what we did there. So that is one thing that this chart shows.
6662 I have to say that when I look at this -- this is why I asked the question of CNOC this morning. I just don't understand why the ISPs are so in favour of 95th percentile. Like I said, a high 95th can be caused by very high usage, it can be caused by moderate use. A low 95th can occur with very low total usage or high total usage.
6663 When you look at this, no one can know who will pay appropriately, who will win, who will lose. Somebody will figure it out. Somebody will, and there will be regulatory arbitrage, and that's another problem.
6664 The third thing from this chart -- and this is directly for Vice Chair Pentefountas, who has asked the question several times today, including of me -- when I say that there is an incentive and an ability with capacity pricing or 95th to drive up your traffic, the question has become: How likely is that.
6665 I look at this and I say: It's possible. It can be done.
6666 I have the actual empirical evidence from the U.K., where they are doing it. They are not doing it for nothing.
6667 But if you look at this chart, it proves that you can get, with a given 95th percentile, low usage or high usage.
6668 So if you go back to my 13,000 95th percentile peak throughput, 13,000 megabytes per second, if you were at 40,000 of total gigs and you saw that your competitor, with the same 95th peak, was at 100,000, it can be done. Somebody is doing it. In fact, many are doing it.
6669 So you will be able to -- you can, and you will -- you will want to drive your traffic to 100, since you are paying the same amount.
6670 I think this charts shows that it can be done and it is being done.
6671 With that, I conclude our remarks and welcome your questions.
6672 THE CHAIRPERSON: Thank you.
6673 As I mentioned, this has been largely caused by Bell because Bell is such a large supplier of wholesale access to ISPs, so we want to make sure that we understand this correctly.
6674 Correct me if I am wrong, but I have the feeling that you are basically saying: Our present costing, tariffs, et cetera, are all based on a certain assumption of how Bell builds its network, and that is based on usage, which you say is not the experience that you have, and that is then reflected in the cost rates.
6675 That is really where the problem is. If you now adopt a system whereby people -- let's say that we adopt the MTS system. Then you really have to have a different costing methodology, because the costing methodology is based on certain assumptions, which, basically, is the normal way that the Bell network is based and they are reflected in this whole costing. Now you suddenly are having usage, which is not reflective of those assumptions. Therefore, we will not recover what is due to us.
6676 MR. BIBIC: That is correct, Mr. Chairman, except that I would correct the very first part of your summary of our position.
6677 We start with the premise, like cable -- and that was the second half of my opening statement here -- that 95th or capacity pricing measures the wrong thing, in the wrong place --
6678 THE CHAIRPERSON: I understand. I wasn't talking about the 95th percentile.
6679 MR. BIBIC: -- and I don't want to abandon that.
6680 Now, if the Commission were, nevertheless, to go with either one of those models, then, absolutely, what we are saying is, it will drive costs in our network. Those costs aren't reflected in the way that cost studies are done today, and if we were to do it, those wholesale rates would either be higher than our retail prices, as you see them, the rack rates, or they would be below, but with a very small margin.
6681 And I say this not to be controversial, or not to be insulting to anybody: I really don't think that the Commission would allow that to happen. Having gone through all of these hearings to establish mandated access, to knock that away by having a rate above retail just isn't going to happen.
6682 THE CHAIRPERSON: Before we get to that -- I am coming to it, but your basic assumption is that Phase II costing is based on the way that Bell builds its network, which is based on certain assumptions of usage, et cetera, and by us allowing the wholesaler to follow the MTS model, that methodology is no longer applicable, because it will drive everybody to maximize rather than what is right now -- I don't know on what it is based, but it is certainly not based on everybody maximizing their usage.
6683 MR. BIBIC: Correct, and that would apply to the MTS model, the Primus model, which is a derivation of MTS, or 95th. Correct.
6684 THE CHAIRPERSON: Now, why do you then assume that we will not -- you say that we will not allow the wholesale cost to be higher than the retail cost. That is perfectly logical, it shouldn't be. It should never be higher. That doesn't make sense.
6685 But, in that case, it means that your costing methodology would be wrong. So, if you changed the costing methodology, surely you could live with the MTS model, too.
6686 You are worried that we will, in effect, misapply the costing, or that we will use the current costing methodology and thereby come up with an incongruous result, and that will artificially reduce the rate, so that the retail is higher than the wholesale.
6687 MR. BIBIC: Correct. I believe that the wholesale rates or costs would be artificially modified to get them below retail.
6688 By the way, on retail -- and Mr. Katz has asked this of a couple of intervenors today. Our retail rates are what they are as a function of the competitive marketplace. The key thing is, if there is extra usage caused by our retail end users, that doesn't mean that we have to go back and redo our costs. In a retail market we actually bill for usage. That's how we recover our cost, we just bill for usage.
6689 We have the flat rate, we have different price points, we have different plans, we have different caps, and we bill for usage.
6690 That's how we do it at retail. But if we are going to adopt cost-based rates for wholesale, then we should adopt cost-based rates, and we should adopt cost-based rates based on the costs that a particular pricing model that the Commission adopts will cause.
6691 But, then, to turn around and artificially modify them because a whole host of small ISPs have come forward and said that they need margin protection -- and, again, I am not making this up. Primus asked for a rate review for TPIA. Mr. Cohen's entire presentation last week was on that. BCBA this morning, or an hour ago, asked for this.
6692 THE CHAIRPERSON: I have no problem with you saying that they will ask for it. I have a bit of a problem with your assumption that we will accede to it and that we will set artificial rates just to produce a difference in the figure. That suggests something different. That implies a bit of a lack of faith in our ability to cost things properly.
6693 But leave that as it be. I just want to understand -- I want to hear, especially, from CNOC and the others.
6694 You have heard Bell, and I think you explained to us this morning, very clearly, that, in effect, it is your lack of faith in the costing methodology -- or that your present costing methodology cannot be used in order for Bell or the MTS model -- the MTS model or the Primus model, without it producing either no margin or, actually, a negative margin between wholesale and retail prices.
6695 That, in essence, is the nub of your issue, isn't it?
6696 MR. BIBIC: No, and I know that I started this morning -- my very first intervention, when CNOC was up, I made this point -- or it may have been my third intervention.
6697 And we are having this dialogue, now, several times about this issue.
6698 THE CHAIRPERSON: Yes.
6699 MR. BIBIC: I just want to make it clear that we, steadfastly, believe that volume pricing is the proper way to address this issue. We find that 95th or MTS or Primus' capacity model is the wrong way to do it.
6700 So it's not a question of, "It would be okay if you got the cost right," it's the wrong way.
6701 And to compound the problem --
6702 THE CHAIRPERSON: Why? Tell me why, because I understand the costing, but I don't understand why it is, per se, the wrong model if you changed the costing methodology.
6703 MR. BIBIC: Because there is no correlation between the peaks at the AHSSPI and what is happening in the aggregation network, which is what Figure 5 of our opening statement shows. Whereas, as Rogers showed, or the cablecos show, traffic is a very good proxy for the peaks that occur in the nodes or the DSLAMs.
6704 That's why.
6705 So it's not just kind of a statement of principle, it is a statement of principle backed up with evidence.
6706 Rogers mapped what is happening at their nodes, we mapped what is happening at the AHSSPI, and we looked at the correlations between 95th peak pricing -- or throughput, rather -- and volume. We have established --
6707 THE CHAIRPERSON: If that's true, why then does Shaw, as they testified here, bill their business partners on the 95th percentile? If it isn't reflective of usage, why would they do that?
6708 MR. BIBIC: Jonathan will step in with a specific answer to that, but what is at stake is much different for us. We have over 100 wholesale ISP customers, representing hundreds of thousands of end users.
6709 MTS can put forward the MTS model --
6710 Last year, when we were at the speed matching decision -- I know what the numbers are now -- they had -- and this was on the record -- four wholesale ISPs, representing less than 1,000 end users.
6711 So that's one answer.
6712 And Shaw did say that volume was preferable, but they will do anything. That's because, I think, there is less at stake for them.
6713 Jonathan can answer more --
6714 THE CHAIRPERSON: No, when Mr. Katz asked them about their business, they said quite clearly: Yes, for business we would use the 95th percentile.
6715 MR. DANIELS: Just to be clear, Bell, too, charges 95th percentile for a service that we provide to some of the ISPs for the backbone. It is designed for a point-to-point service, it's not designed for -- as the term that I like, what Rogers used earlier today, or that cable used earlier today -- a distribution aggregated network.
6716 What I heard from Shaw were two things. One, yes, they do that today, as do we, but it is a different service. You don't have the same issues that we are talking about here.
6717 Second of all, in terms of why they say it can be acceptable to them, I guess they don't have the same concerns on their impact of the network that we do, because it's a volume issue.
6718 This matters to us. It matters to us in how we invest on a day-to-day basis.
6719 In the same way, I don't think -- I don't mean to disparage anyone else, but I think we heard from TELUS, for example, that it just doesn't have the same impact on their network.
6720 THE CHAIRPERSON: I understand that.
6721 I see two flags up. First, CNOC.
6722 MR. TACIT: Thank you very much, Mr. Chair.
6723 I would like to make a few comments. First of all, with regards to the chart, it certainly proves that they can --
6724 THE CHAIRPERSON: Which chart?
6725 MR. TACIT: Chart No. 5.
6726 It certainly demonstrates that they can measure 95th percentile.
6727 MR. TACIT: Second, correlating volume and peak isn't the issue. That is exactly the reason we don't want volume charges, is because the correlation is poor. What matters is correlation between peak and costs, and we have said -- and the way networks are built -- that because there is correlation is -- the way it is charged is, according to the 95th percentile method, precisely for that reason.
6728 Now, I found it interesting, you know, we sat around and it turns out that five of the six ISPs that were before you today were around -- have a very similar kind of peak profile and, in fact, it reminded me of an interrogatory response we gave where we basically were asked by a cable carrier, "Here's our profile; would you say that this typical?"
6729 And we said, "Yeah, it's pretty typical". So a lot of ISPs who are in a similar kind of business do have a very similar kind of profile, and to the extent they don't and others have a different profile, it actually is a benefit for the incumbent because they basically get to fill in peaks and charge more without actually having to build the extra capacity.
6730 The other thing that I keep finding curious is this reluctance to build out the network that the incumbents keep talking about.
6731 You know, if we go to this model, even if we recover all the costs, there's going to be more demand on our network and we're going to have to build more.
6732 Well, yeah, that's what we want. That is exactly what competition is supposed to do. And I found it very interesting that the examples that Mr. Bibic just gave were all from the UK, whereas we know we have some functional separation and maybe what that shows is that, in that kind of environment, competition actually works very well and you get a wide variety of consumer choices in packaging.
6733 And finally, I did a bit of research of my own, but was reluctant to introduce this because it seemed like new evidence. But since Mr. Bibic took the liberty of introducing the UK example, it turns out that in Belgium there is an incumbent called Belgacom which offers a DSL service.
6734 And basically, you can get that service; it includes the DSL port, and with it you can buy an Ethernet service that's priced according to the peak capacity model.
6735 And I'm told by one of our European contacts that the price for that, I think, is in the order of below two Euros per megabit per second.
6736 So we're not reinventing the wheel here, and this isn't rocket science.
6737 MR. BIBIC: So may I respond to some of that?
6738 THE CHAIRPERSON: Well, there's another flag up. Let me hear from the other, then -- yeah, go ahead.
6739 MR. MOORE: Hi, Andrew Moore.
6740 Just a simple question. We've heard Bell repeatedly say that 95th percentile measures the wrong thing at the wrong place, yet you can also -- what's stopping Bell basically from also measuring usage at the POI?
6741 Isn't that measuring at the same -- at the wrong place, too? Because from my point of view, you measure either at the customer any point across the network unless you have a high amount of packet loss, the number you're going to get at the end of the month is exactly the same regardless of where you're going to measure the network, so why, compared to 95th percentile, is the -- well, what I'm going to call the revised volume pricing, or RVP, is any better than the 95th percentile in that -- on that aspect.
6742 THE CHAIRPERSON: Mr. Bibic?
6743 MR. BIBIC: So I guess I'll start first with some points to respond to Mr. Tacit.
6744 First, this is minor. This chart at Figure 5 and the data points, those were pulled manually, all right, so the point about we can do this, it was pulled manually over the course of the last week when -- after Vice-Chairman Katz asked the question.
6745 I find it rather ironic, my second point, rather ironic that Mr. Tacit would accuse Bell of being reluctant to build out networks.
6746 If you look at Figure 1 of our first written submission in this proceeding, you'll see that we reproduced a chart by IDC Canada showing the total CAP-X spending by CLECS in Canada in 2010 at 159 million and you have in confidential submissions of ours how much we spend per year just on capacity augmentation; nothing more.
6747 Not roll-outs of fibre to other areas, to new areas, just on capacity augmentation. The 159 is bigger, but not that much bigger, okay. That's the second point.
6748 I mean, it's ironic. I think Commissioner Patrone asked Primus -- and again, I'm not picking on Primus because I think Primus did a very good job last week and they gave very straight answers, but they did say, I think in response to Commissioner Patrone, that they lease all of their networks.
6749 And as far as the UK and my evidence there, I do believe that the UK has 95th percentile too, so I can easily say that 95th percentile and, guess what, you've got midnight madness sales because you have an incentive to use what you're paying for.
6750 And it just makes sense if you're paying for -- any ISP would do that.
6751 I forgot the last question, Mr. Moore's question, but Jonathan, if you have an answer, go ahead.
6752 THE CHAIRPERSON: Mr. Moore, do you want to repeat what you said?
6753 MR. MOORE: Yeah. In short form, we've heard you say multiple times that 95th percentile measures the wrong thing at -- also at the wrong place, yet I don't see the difference in your revised volume pricing model, that basically you can measure usage at any point in the network; you're supposed to get the exact same number at the end unless you have a high amount of packet loss, so how does it measure the right -- like at the right place compared to 95th percentile model?
6754 You could also measure usage at the POI, so I don't see the difference from my point of view as of the location of the measuring network.
6755 MR. DANIELS: It's Jonathan Daniels for Bell.
6756 I think the answer to that question is that we're talking about measuring different things. That's the issue.
6757 So when we say 95th percentile measures the wrong thing at the wrong place, what we're talking about is theoretically -- and I think we had this discussion last week, so I won't repeat the whole notion. But theoretically, if we could measure 95th percentile on all of our congestion points 1, 2 and 3 and then somehow figure out what is retail and what each -- not collectively, but what each ISP is responsible for and their portion, and theoretically, 95th percentile measured at that point would actually be the right place to measure it.
6758 But you can't do that. No one's suggesting that you can do that, I don't -- no one in this room.
6759 And so when we say you can't -- that 95th percentile is measuring the wrong thing, it's because it's measuring and it goes back to the correlation issue that Mr. Bibic talked about.
6760 Our AVP, it's true, it doesn't matter in terms of not fussing about where it's measured; it's the point it's measuring the right thing because it's measuring something that's correlated to actually what affects our costs on our network.
6761 THE CHAIRPERSON: Okay. I see the flags. Hang on, hang on. One second here.
6762 So but -- so you agree, then, with what Mr. Engelhart said. You were looking for a proxy. What would be ideal if measuring at all these points and the closest proxy and the one that's most representative is volume based.
6763 MR. DANIELS: Absolutely, I agree with Mr. Engelhart on that.
6764 THE CHAIRPERSON: Okay. Thank you.
6765 Okay. Let me go over these flags here. Who's in the middle there? Go ahead, Shaw.
6766 MR. BRAZEAU: Just to clarify a few points, Mr. Chairman. Jean Brazeau from Shaw.
6767 We agree that -- we still agree that usage and peak provides the best correlation. The only point we were making is that if you look at -- you can find a correlation between traffic and usage and which would allow us to do the 95 percentile billing.
6768 But the best correlation is certainly the one that's been proposed by Bell and the cablecos, and the ones that we also think best suits Shaw, and it's cable network.
6769 As for the 95th percentile, I'll let Mr. McEwen address that point.
6770 MR. McEWEN: Yes. We wanted to just comment on that today we do bill 95th percentile for ISPs, but that is not linked to the provisioning systems that are used to provision TPIA, so it's not just an issue of measuring; it's not just an issue of billing. The -- those have to be linked to the provisioning systems as well.
6771 MR. BRAZEAU: And also, it's really our business wholesale services, and we're really talking about when we're selling fibre to large users, that's when we use the 95 percentile.
6772 We don't use it for any of our TPIA services.
6773 THE CHAIRPERSON: Thank you.
6774 On this side here, gentleman at the very back.
6775 MR. GAUDRAULT: It's Marc, from TekSavvy.
6776 Just anecdotally, I just don't buy it, you know. Like TekSavvy, I mean, we've been sort of in the mix of all this and our peak usage is between 10:00 p.m. and 11:00 p.m. And I'm pretty sure if you look around the room, just about all of these incumbents, it's -- that's the peak peak all of their networks.
6777 So like that we're not measuring it at the right time, I just don't buy it. It just doesn't pass the smell test.
6778 THE CHAIRPERSON: Okay. Over there.
6779 The gentleman behind Mr. Rogers, whoever he is. Yeah, Mr. Cohen.
6780 MR. COHEN: Mr. Chairman, when we talk about these midnight madness sales, this Commission is being asked to accept the perverse notion that independent ISPs making fuller use of a capacity-bound network is a bad thing.
6781 It's not a bad thing; it's a good thing. It's more efficient use of networks.
6782 The more gigabytes that could be transferred between users and their destinations for the same amount of network capacity, the more efficient the utilization, the better it is for everyone.
6783 So this -- I would really like to see this nonsense being stopped about the fact that we're going to stimulate more usage in any other context, whether it be hydro or -- you know, like it would be seen as a better thing to make more even flat distribution of the capacity and optimize its usage.
6784 So let's turn to the -- just question that Bell has raised about the fact that they would not be compensated for their usage downstream in their network because, in fact, we would be paying for peak capacity at the wrong time.
6785 If I look at their Figure 4 and I look at the top chart for 8:00 p.m. and I look at the bottom chart for 1:00 a.m. and then I think about how a cost study is done, the cost of augmenting the 250 megabit per second link in Chart 1 -- sorry, in the bottom part, the 1:00 a.m., the 400 megabits, that cost of augmenting that link would be in the numerator of their cost study, if you like.
6786 They are going to have to calculate that cost in their calculations when they do their cost study just like they would have to calculate a different link's peak at a different time. But all the costs of augmentation will go in to this study, so then the question is, what is the denominator. What do we divide by?
6787 Well, yes, it's true that this particular ISP's contribution to the denominator would only be 600 megabits per second and every ISP's peak utilization would be added together to get into that denominator until we took the sum of all the peaks of the ISPs.
6788 And then what we would end up with is a cost of aggregation across all of the network divided by the total of the ISP's peaks.
6789 So to then say that we would not be compensating Bell or Bell would not be getting any remuneration for their augmentation at this 1:00 a.m. is a fallacy.
6790 Every measurement, every unit of measure that you use in the denominator will take the costs of development and spread them over that unit of measure. And if the relationship between peak traffic and total gigabytes transferred through a month were constant, then it wouldn't matter.
6791 The problem is, what happens as that relationship changes? And this is where I think we're seeing something somewhat disingenuous in this proceeding.
6792 What we have heard from Telus is that traffic is growing at 30 to 40 percent per year. We have also heard from Telus that the traffic pattern is flattening as times goes by, so -- and I asked them. Their 30, 40 percent they were referring to was total gigabytes transferred, not peak utilization.
6793 So in your mind, picture what's going to happen from this year to next year. Total gigabytes transferred will grow by 40 percent, but peak utilization will not grow by 40 percent because the traffic is flattening over time.
6794 Now, I'm going to be paying a rate that's based on where that relationship was today, so when total gigabytes transferred grows by 40 percent, my costs will grow from 40 percent. But because traffic has flattened, the costs of building the network have not grown by 40 percent.
6795 And that's -- the key here is which measure more accurately will track what happens to the underlying costs.
6796 Thank you.
6797 THE CHAIRPERSON: Mr. Bibic, there's some strong language here about nonsense and disingenuousness and so on. Maybe you want to respond.
6798 MR. BIBC: Yes, there's -- I think there's three basic points here.
6799 I think the nonsense point relates to somehow what folks are associating me with is saying that we want to stop usage, which is not true.
6800 I mean, paragraph 12 of this rebuttal statement says there is absolutely nothing wrong with more usage of the Internet, but those who use more should pay more for their usage.
6801 I just want to be categorically clear on it.
6802 It's fine with us. It's great if people use the Internet more. We just want to make sure that those who use more should pay more. That's one.
6803 Two, Mr. Cohen said that, well, if he's going to pay for something, you know, he should make the most efficient use of what he's paying for. I think I said that about three times in the last half hour.
6804 I agree. That makes perfect sense to me. And I think that's -- but that's going to cause extra costs in the aggregation network.
6805 And on that, which is Mr. Cohen's third point, on the costing, I'll turn it over to Jonathan.
6806 MR. DANIELS: So as I listened to Mr. Cohen's description, I think what will help to understand is if you go back -- and we really simplified things, I grant you. It's really hard to explain costing with a diagram of what's going on the network.
6807 But if you look at Figure 4 at the top there, in this simplified notion we said that the links are built to their maximum capacity for the peak period, which is the 250, the 100, the 50, the 200.
6808 Now, I have to admit to you that's a little misleading in the sense of, of course there's retail use on top of that, what that actual link would be at. But for simplicity's sake, we then said, but it turns out the bottom, the person is jumping up to 500 megabits here at Cluster A from the central office.
6809 The users there are causing that to jump up and put that linkage into red.
6810 And our whole point, and I think you understood this before, Mr. Chair, when you asked the question, our whole point is that 95th percentile will result in more usage, therefore, more costs which aren't reflected in the cost model.
6811 And then why do I say that they're not reflected in our cost model? Because our cost model works on the assumption today -- let me take a step back -- at how we actually provision today based on the use we see today, based on the usage patterns that were experienced today jointly between wholesale and retail. We consider it on an aggregate.
6812 So when the Commission directed us to file a cost study, we filed a cost study with the set now of the mark-up that's been set by the -- we're saying the cost-based mark-up using the same one for access.
6813 We filed an AVP rate. The new rate would be 17.8 cents.
6814 The Commission asked us, "Give us the rate for 95th percentile". Using the exact same approach, the same -- we took that same cost. We end up with a rate for 95th percentile per megabit per second, which is the $39 that Mr. Bibic referred to, 39.50.
6815 So those two costs, underlying it out of the model, are exactly the same. And you put them together; therefore, a revenue that we will get will be exactly the same. But the difference is that our cost model assumes that its usage generally as we see today in our network. And that's the difference, that the bottom there is meant to say that we think 95th percentile and even more so with MTF's proposal or the Primus proposal, that it will cause costs to go up even greater than what are captured by our cost study.
6816 And I think you understood that from earlier, but that's to respond to Mr. Cohen.
6817 THE CHAIRPERSON: Thank you.
6818 Mr. Bhullar?
6819 MS GRIFFIN-MUIR: I think you're referring to me.
6820 Oh, come on.
6821 MR. BHULLAR: Thank you, Mr. Chairman.
6822 I want to make one observation on Figure 1, if we can look at it.
6823 It is shown as that No. 1 to the left is where congestion happened and, according to Bell, it appears that somehow the congestion is going to be managed through usage-based billing.
6824 Let's look at it, how a congestion could happen.
6825 Say this link was -- or the capacity coming out was 600 megabits per second. A grandma dials in while the network is just close to 600 megabits. She happens to want to watch a religious program, and the congestion will happen. But that grandma, on average, will not be paying usage-based billing extra charges for that month.
6826 So there is a basic disconnect to assume that somehow congestion and specific users' behaviour are directly related.
6827 Like if somebody is using capacity as has been flagged before on a pattern that is more conducive to the overall pattern, like it fills in the gaps, they're not causing congestion, but they'll be paying UBB.
6828 So the issue is, I do not believe that for innovation you want to be forcing everybody to look like mini Bell or a duopoly, so that is Point 1.
6829 And the other part is that they are flagged a lot of the ISP traffic patterns. I just very recently changed to TekSavvy, and we know very well that these are new entrants to many territories.
6830 So as a customer, a lot of the customers are upgrading. When they move to the new ISP, they are TekSavvy people, no pun intended, so they move to the higher band width.
6831 So has Bell caught up to that change when they turn up these new POIs or access points? We do not know. But they are definitely getting more money for the wholesale rates from TekSavvy.
6832 So there is a disconnect to -- like this is not a mature market where TekSavvy is everywhere within Bell Canada in every city. So once that happens, you will see that their customers are going to be no different than Bell's customers. And all this talk about, you know, congestion at Point 2, congestion at Point 3, that's nebulous.
6833 So the third point is, like they are offloading from this network to IPTV. Once they launch it, this is going to be an under-utilized network unless they pare it down to big congestion.
6834 So we have to really watch what is happening that, you know, like we are getting a lot of these ---
6835 THE CHAIRPERSON: You're on to a different point here. Let's stick to the point we are.
6836 Okay. Rogers.
6837 MR. ENGELHART: Thank you, Mr. Chair.
6838 I just wanted to add a point to really the discussion you were having with Bell.
6839 So the way these costing models work, you have the cost of the network. That goes in the numerator. And then you divide that by a unit, so the usage-based costs or the traffic-sensitive costs divided by a unit. That unit could be volume; it could be speed at the peak; it should be shoe size; it could be your number of cats.
6840 It's -- there's some unit that you hope will approximate the cost driver.
6841 So first of all, it's not true, as CNOC keep saying, that by using the 95th percentile measure they're paying us for the whole day, whether they use it or not. It's just not true.
6842 That would only be true if we had the cost of the whole day in the numerator. Dividing it by a different denominator doesn't magically mean that they're paying us for the whole day, so that's point No. 1.
6843 Point No. 2, if you accept that the speed at the peak is not as directly related to the costs as volume, which is what we're arguing and Bell is arguing, there are two problems.
6844 Problem 1, as you've indicated in your questions to Bell and as Bell has agreed, is that the distribution of traffic might change; in other words, the numerator might get bigger as a result of your pricing mechanism, and that's going to mess things up.
6845 But the other problem is that even if that traffic distribution doesn't change, over time there will be a growth in the Internet. There will be a growth in volume. There will be a growth in the underlying cost drivers.
6846 And if your unit of measurement, the denominator, isn't tracking that carefully, then, as time goes by, you will be either under-compensated or over-compensated.
6847 So even if the traffic distribution doesn't change you still want that denominator to capture the cost drivers as closely as possible.
6848 THE CHAIRPERSON: Okay.
6849 Mr. Tacit?
6850 MR. TACIT: It's Chris Tacit again, for CNOC.
6851 On the costing issue there has been a counterbalancing factor and that is that because there are so few costing reviews over the years, what has in fact actually happened is that wholesale rates have become less and less competitive over time so to the extent that because -- as actual costs in the industry have declined, rates haven't fallen to reflect that and the implicit mark-ups have increased.
6852 One of the ways to deal with this whole issue of making sure that we have the proper cost drivers and also to ensure that we don't start inflating those rates simply through the passage of time, is to have more frequent reviews that actually look at what is happening in the marketplace with regards to transport and access costs.
6853 I think if the Commission were to do that a lot of these problems would go away. Frankly, if it doesn't do that, then regardless of whichever model it adopts, there are going to be problems. I think this is a fundamental point, you know, that needs to be conveyed.
6854 THE CHAIRPERSON: Do you want another Phase II costing exercise?
6855 MR. TACIT: No, I don't want a Phase II costing exercise. We just want the actual cost studies to be updated more frequently. We are not suggesting a review of the methodology. We are just saying that we need to have the costs reviewed more often.
6856 The fact is this is a rapidly changing industry and in order for a regulation to be official --
6857 THE CHAIRPERSON: Mr. Tacit, I got your point. I made a joke.
6858 MR. TACIT: Yeah, okay.
6859 MR. TACIT: Excellent.
6860 THE CHAIRPERSON: TELUS...?
6861 MR. ROMANIUK: Thank you. It's Orest Romaniuk for TELUS; two points.
6862 So first off, as we indicated, costs are not going down. Yes, the electronic component is, but anywhere from 35 to 85 percent increase in the other factors as you push the facilities out, those shovels are getting more and more expensive to do the digging. So it is very expensive.
6863 Secondly, the point earlier on that was made that an increase of 40 percent in volume doesn't necessitate a cost increase. I disagree with that totally because what is happening, your 40 percent increase is -- it may not increase the cost in the first year but it increases your base and when your peak gets higher and higher to where you are touching the maximum capacity.
6864 Next year, yeah, on a 40 more percent of base all of a sudden your peak is over the capacity of the network such that you have to do a rebuild.
6865 So absolutely, the 40 percent it just keeps on pushing it up and then it peaks or causing you to rebuild. That was the point that I don't think was being captured earlier on.
6866 Thank you.
6867 THE CHAIRPERSON: Okay, the gentleman --
6868 MR. SANDIFORD: Yeah, I have a couple. It's Bill Sandiford for CNOC.
6869 I have a couple of comments with regards to some of the things that were said by Mr. Bibic. He mentioned that he was glad that they agreed with Distributel's assertion that it's a good thing that we want to officially utilize the network.
6870 What they don't realize is that their AVP model does not incent ISPs to officially utilize the network. If we are forced to pay for a byte of data, regardless of when it's transferred, we are going to have no incentives whatsoever to flatten out those peaks and use our links more effectively.
6871 The second thing is from -- he made reference to his own comments in paragraph 12 of their comment where they see nothing wrong with using more.
6872 I kind of believe that that's -- a more accurate statement would be they see nothing wrong with using more as long as they can charge you for it. And they want to charge for each and every single byte that traverses the network and they want to do that regardless of the cost impact of that byte.
6873 None of these principles are sound in trying to officially utilize the network capacity that we are willing to pay for.
6874 THE CHAIRPERSON: Mr. Bibic, you have mentioned that you were going to charge at a rate of 1.78 a gigabyte. What happened to all this talk we heard last week about different rates of access and usage in one combined rate and that's it --
6875 MR. BIBIC: No, no.
6876 THE CHAIRPERSON: -- or we are missing something here?
6877 MR. BIBIC: Yes, yes. Okay. So let's go through that again.
6878 There should be an access rate with the mark-ups that the Commission has already approved --
6879 THE CHAIRPERSON: Okay.
6880 MR. BIBIC: -- in the past for access.
6881 On top of that a usage rate based on AVP with that usage rate having a mark-up that's the same as for access.
6882 THE CHAIRPERSON: And the access rate would vary according to speed?
6883 MR. BIBIC: Absolutely. Now, with that --
6884 THE CHAIRPERSON: To the extent that the speed has a cost -- is a cost driver?
6885 MR. BIBIC: Correct. And then if it's one rate -- and this is one of the problems with -- one of the problems with MTS.
6886 If there is one rate for any speed at all that you can drive through that network now, the arbitrage gets massive and it's going to significantly impact us in the retail market and cause a big problem in the incentives to invest which is exactly the ground we covered last year in the hearing.
6887 THE CHAIRPERSON: Yes.
6888 Now when -- after your appearance, several people commented on your Chart 3 and I believe Jean-François Mezei in particular from Vaxination informatique was particularly critical and said this really doesn't represent -- your IPTV is not a separate -- and I forgot his exact criticism. But he suggested that it is joined for part of the BAS and that the way you characterize it here as being totally separate and one having no impact on the other was not a correct representation.
6889 Do you remember what Mr. Jean-François Mezei said?
6890 Is he here? Can you repeat what you had to say about that?
6891 MR. MEZEI: In the original Bell filings some time ago the graphs were showing a shared path between the DSLAM and let's say the downtown core where the BAS would be. And so the indication was that the IPTV and Internet were sharing part of the way from the DSLAM in the suburb to somewhere downtown.
6892 Bell had made it very clear that the IPTV would split off from the Internet before reaching the BAS because it went into its own centre where -- one video server form in the centre of the city and that they didn't share the BAS.
6893 So based on those original filings my assumption was that it did share the BAS. Now, today or yesterday, I believe, Bell mentioned that they are split up by the DSLAM so I don't know what to believe anymore.
6894 THE CHAIRPERSON: Bell, do you want to comment on this?
6895 MR. CONDON: Yes, Mr. Chairman.
6896 What we have said consistently -- we said this last year in the ADSL -- in the matching speed hearings, sorry -- and again this year, that they do not share that path.
6897 There is no smoking gun here, Commissioner. So there is no shared network. It does not impact our congestion. TV does not impact congestion.
6898 MR. BIBIC: Mr. Chairman, may I go back and just quickly respond to Mr. Sandiford?
6899 THE CHAIRPERSON: Sure.
6900 MR. BIBIC: So a couple of things.
6901 One is the AVP rate or the volume rate we are now proposing is cost-based. So if it's cost-based and it's done properly it shouldn't have any impacts on the incentives to use the network.
6902 Two, what the ISPs under AVP would essentially be doing is they are paying based on volume including -- and there is breakage involved there. They can have an unlimited price and price it as such. People, if they use only 5 gigs, it's going to have very little impact on their AVP costs.
6903 Look, in the retail market Bell has usage-based billing once you hit your data caps. Rogers does. Videotron does.
6904 This is the bulk of users. Probably greater than 50 percent of Canadian users are in Ontario and Quebec just given the size of those markets. Growth is dramatic.
6905 It's only those who are using the absolute most who are paying the most and that's the way it should be. People are accustomed to it. You see it with fuel. You see it with electricity. You see it with water consumption. You see it with wireless. You see it with long distance.
6906 THE CHAIRPERSON: When we originally called this hearing after all the uproar, et cetera the uproar was number one because you were actually imposing on the ISP to become mini-Bells that they would have to impose the same caps that you have got. All of that, as far as I can see under this proposal of yours, is gone.
6907 And the second part we wanted to make sure that the ISPs had an ability to differentiate the product and are free to innovate and develop the product without being restrained by you. And you are saying this does not restrain them in any way to do that, if I understand you correctly.
6908 MR. BIBIC: That is absolutely right.
6909 So the first point, it's gone. If an ISP wants to offer unlimited and play the averages, recognizing there will be very high users, they can subsidize that with their very low users, they are free to do it. No one is holding them to our pricing plans.
6910 In terms of innovating and creating new products, CNOC in the presentation they gave you in January/February had a whole bunch of things that -- a whole bunch of innovations that they say they brought onto the market. As an aside we do those things too but that's neither here nor there. Those were being done in the current environment and nothing will change.
6911 We are respecting with AVP the speed matching decision. We are respecting mandated access.
6912 In fact, you know, we live in a world of mandated access. We would prefer to live in a world of commercial negotiations.
6913 There are a whole host of things that we do in the marketplace at wholesale today, given that we live in this world, to encourage ISPs to come on our network and not the cable network, et cetera. We will continue to do those things.
6914 THE CHAIRPERSON: But I mean, I recall too you saying on Monday that you for your own Internet customer you will continue to have a regime with caps. So from which I take it an ISP buying from your wholesaler can say, "Here, come to me. Bell caps you at 60 or whatever it is. With me you have unlimited access and my rate is more competitive".
6915 MR. BIBIC: That is precisely what they will do, and that's fine.
6916 THE CHAIRPERSON: Okay, thank you.
6918 MR. TACIT: Yes. So just for a moment looking at the IPTV charge, just going back to that issue for a minute, the thing that puzzles us is the notion that somehow that traffic would be on separate fibre links or fibres.
6919 It seems to us that it's much more likely that there would be a single set of fibres transporting all of that traffic. It's just hard for us to get our mind wrapped around.
6920 Would they really lay parallel fibres or take different rights of way just because it's IPTV traffic?
6921 The other question that I have is --
6922 THE CHAIRPERSON: Well, hang on, before you get to the next one, let's have an answer to this one.
6923 MR. CONDON: Commissioners, I can only repeat. It may not be on separate fibres. It may be on separate wavelengths in the network. It may be on separate fibres in the same conduit. It usually, probably is. I don't know.
6924 All I can tell you is that there is -- they do not share the same pipe. I don't know what more I can say to convince --
6925 THE CHAIRPERSON: Are you speaking metaphorically because it was the same pipe -- it was the same fibres. It's probably through the same pipe, you know.
6926 MR. CONDON: Or the same -- what is the word I want to --
6927 MR. BIBIC: If I may, actually, I think, Mr. Chairman, you expressed this whole issue the clearest when you appeared in front of Industry Committee in February.
6928 You said to the members of that committee "With IPTV basically there is a pipe and it is divided into various sections. There is one section that is public Internet and everything and there is another section that provides the specific service IPTV and that's what cable does." That's what Carl is saying right now, exactly how you expressed it to committee.
6929 THE CHAIRPERSON: I got something right, thank you.
6930 THE CHAIRPERSON: Okay.
6931 CNOC, is that your question?
6932 MR. TACIT: Okay. So following up on that, then there is this interesting issue that, you know, they have to -- if you have a single phone line you have to use one modem for both the high speed and IPTV signals presumably because the IPTV signal rides along with the high speed at some point.
6933 So again, it's just there is a lack of clarity. I guess what we would suggest is that perhaps the Commission can dig into this issue a little bit more before this whole proceeding is wrapping up.
6934 MR. CONDON: Commissioner --
6935 THE CHAIRPERSON: That's why we are having this hearing here. If you have got a question you just pose it to him.
6936 I want Bell to answer. I'm not proposing to "dig into it". I mean I have gone so far on the assumption that this is correct and I even testified before Parliament on that basis.
6937 So please answer their question.
6938 MR. BIBIC: Before he does, and if you want the answer we will give the answer, but this whole issue of, you know, one modem, two modems, getting access to IPTV and all that, that was debated at length at the last hearing and the Commission's ruling on that was clear. And this was such a big issue for us in terms of access to IPTV.
6939 This is not what this hearing is about. I feel pretty strongly about it. It's clearly out of scope. I don't think it should be answered, frankly. That was debated, discussed and decided.
6940 THE CHAIRPERSON: Mr. Bibic, the only reason I am asking is people are allegedly trying to undermine your credibility saying that Shaw doesn't properly represent it and you are telling me it does and I am hearing the questions again. It still doesn't.
6941 So that's why I wanted -- not because we are going to make a ruling on this point.
6942 MR. BIBIC: Okay, that's fine.
6943 Mr. Tacit moved on to a different point though.
6944 MR. CONDON: It's well documented. I don't know why anybody is surprised.
6945 But we do use the VDSL2 protocol for both the Internet service and the TV service. That protocol runs from the central office to the customer's home. So it does not run on any part of a shared network. It is a point-to-point network between the customer's home and the central office.
6946 So it doesn't cause -- we talked about this before, that there is a certain amount of bandwidth that the customer can choose to use for watching TV or using for Internet but it doesn't cause congestion or impact his neighbour at all.
6947 THE CHAIRPERSON: Thank you. Okay.
6948 Further comments -- is it Mr. Cohen(ph) or stand up, please. I can't see.
6949 MR. MEZEI: Jean-François Mezei.
6950 THE CHAIRPERSON: Okay.
6951 MR. MEZEI: I would like Bell maybe to confirm a statement or deny it that IPTV and Internet would physically travel on the same pipe but would be logically separated into separate VLANs that each would have allocated bandwidth so they don't play within there. Is that a correct assumption, yes or no?
6952 MR. CONDON: Just further to what I just said, it's not correct. They each have their own VLANs. As I said, they travel on the pipe, the point-to-point pipe between the central office and the customer's home.
6953 The IPTV has priority in that pipe but the VLAN -- if it's not being used the customer's Internet service could use that VLAN, that capacity. This is the same as we said last year at this time.
6954 THE CHAIRPERSON: CNOC.
6955 MR. SANDIFORD: It's Bill Sandiford here.
6956 I think to put this issue to rest perhaps once and for all, I think what we are saying where we are differing in opinion from what Bell is saying, it's nice to have -- be presented with a diagram that shows a bunch of clouds that get separated and whatnot, but when we build networks we need to think about all aspects of the networks. We need to think about the rights of ways. We need to think about the conduits. We need to think about the fibre cables, et cetera.
6957 So for Bell to come to us with the assertion of this is a totally separate network; it's not a shared component and whatnot, sure, maybe they separate out the VLAN and they put it on a different fibre with a different light pulse but there are so many components of that network, the things that I just said, the actual fibre cables, the pole rights of ways and they are all the same.
6958 I mean if they need to deliver IPTV service from Toronto to London, Ontario they are not going to use a different right of way. They are not going to use different fibres. They are not going to use different conduits. They may not even use different optical here. We just heard Mr. Condon say it might even be on a different wavelength in the same fibre.
6959 So the assertion that these things aren't shared and it isn't the same network sort of seems odd to us. And I think that's where we are getting it from. There are so many shared elements of this. We can separate it by putting pretty diagrams of clouds on it, but I think that's it.
6960 THE CHAIRPERSON: Okay, Mr. Daniels, the last point on this issue.
6961 MR. DANIELS: Just on this issue. So I think we have talked about there is the difference between to the end-user and from the node.
6962 But in terms of the backhaul that we are talking about, Mr. Condon was correct in terms of describing how it's on a separate network. It sometimes may be on the same fibre, depending where it is going.
6963 But the video network is separate. The locations, all of our Internet go back to broadband access servers. Those are not necessarily the same locations where we are backhauling for the video.
6964 So physically it sometimes may be on the same fibre because it may be the same and often it's not. That's why it is completely separate networks.
6965 THE CHAIRPERSON: Okay. Thank you.
6966 Len, you had some questions?
6967 COMMISSIONER KATZ: Yes, thank you.
6968 Good afternoon and a long afternoon it's been.
6969 I am going to ask you the same question I asked Shaw this morning: Do you recover all your costs today at the retail level for broadband services?
6970 MR. BIBIC: Well, Mr. Vice Chairman Katz, if you take a look at paragraph three, unfortunately our capital-intensity ratio is greater than 100 percent.
6971 COMMISSIONER KATZ: That's your capital intensive ratio.
6972 MR. BIBIC: That's correct.
6973 COMMSISIONER KATZ: Do you recover all your costs? Are you in the business of making money? When you price your products do you price your products with the intent of generating a return to your shareholders?
6974 MR. BIBIC: That's the ultimate goal.
6975 COMMISSIONER KATZ: Okay. So the price is priced to that market and to that goal?
6976 MR. BIBIC: That's the ultimate goal. And in the retail market -- so our first point.
6977 We are in a world right now, given the intense capital expenditures we made over the last five years, to catch up with cable with a billion dollar-plus CAPEX program for FTTN and FTTH in our territory in a position where today CAPEX, just CAPEX alone -- I'm not talking about OPEX -- CAPEX alone exceeds our Internet revenues.
6978 But over time you hope to recover that and generate a return on that investment. How do we do that in the retail market? Different -- we have tiered pricing, different price points accommodating different users with different bandwidth caps ensuring that those who use more pay more. That's how we approached the problem on the retail side.
6979 On the wholesale side there is the world's biggest loophole. Cable has had the authority to do UBB at wholesale for over 10 years and has actually implemented it.
6980 We have all the -- well, I'm exaggerating. We have most of the wholesale end users and we are not in a position to recover the disproportionate share of the traffic caused by the wholesale end-users. That's all we are trying to do here.
6981 And we have stepped away from UBB and we have AVP and we think it does -- it will do a fine job of closing that loophole, will be completely fair, will be a good economic ITMP and will give ISPs the full flexibility to do what they want to do. But that's how we recover on the retail side, tiered pricing with UBB.
6982 COMMISSIONER KATZ: So if we take how you recover on the retail side and we subtract from that those services that you no longer have to provide from a wholesale perspective, that being billing, customer service, sales and marketing and whatever commissions go into that as well, and whatever network costs you may save by virtue of the fact that the wholesale regime may have some infrastructure as well, would it not be reasonable to assume that the price at the wholesale level should be lower than the price at the retail level?
6983 MR. BIBIC: I have got two answers to that.
6984 One, you have to add back the costs that will be caused by -- depending on the pricing model you adopt, if it's capacity pricing MTS model or 95th percentile it will cause more costs in the aggregation network which is the crux of our case and, I believe, cable's case. So I would back those costs first and then we can have a discussion.
6985 But what you indicated sounds an awful lot to me like retail-minus which is what our wholesale UBB proposal which you approved, it was retail-minus. It was our retail UBB rate less 15 percent. That unfortunately was rejected and here is where we are today.
6986 COMMISSIONER KATZ: But I am trying to understand whether retail-minus all those things can equate to cost-plus and all as a matter of doing, is solving for a couple of variables.
6987 Because at the end of the day, if you are making profit on your retail side and we just take off all those costs and, granted, add back whatever additional costs are driven by this, then the question only becomes is there sufficient margin for independent ISPs to take that, add back their billing, their customer service, their network costs, their sales and marketing costs and provide a competitive service to the marketplace.
6988 MR. BIBIC: Vice Chairman Katz, you are making a huge assumption that there is profit right now in the Internet business. I mean, you know full well from the early days of wireless when you were in the industry that -- and the Rogers folks can attest to this -- it took a very, very long time to get that business into a level of profitability.
6989 We are in a stage of incredible capital expenditures. The number is there. CAPEX is --
6990 COMMISSIONER KATZ: That's why I asked you whether you were making money today.
6991 MR. BIBIC: CAPEX -- I mean the goal, and I said the goal obviously is to do that. Now, of course it's a long term goal and you make your investments.
6992 You expect to reach -- to generate a return on investment over time. That was why we were so adamant last year that you -- you know, you have to be careful in how you mandate access and shield IPTV, et cetera, because the margins are so thin when they eventually hit that, you know, any mandated access done the wrong way especially is going to diminish those incentives to invest.
6993 So I mean you are asking me about retail-minus. I have a lot of problems with the question because you are asking me about whether retail-minus ought to work. That was rejected unfortunately.
6994 Built into your question is an assumption that we are making profits today. You see paragraph three. Obviously, we are investing to stay in the game to keep up with cable and hopefully in the long run if we win --
6995 COMMISSIONER KATZ: Capital --
6996 MR. BIBIC: -- the broadband home, generate a return on investment.
6997 COMMISSIONER KATZ: Capital investment though is not an indicator of profitability and success. It's return on investment. It's your business model. It's a five year or seven year or 10 year, or whatever investment model you have chosen to take.
6998 But I can only assume the marketplace is competitive. You folks and whoever you are competing against on the facility side or anybody else are all there to provide a return to your shareholders.
6999 So I can only assume that, based on your business case and whatever years you assume your business model is you are all in the business of making money. So you are all profitable or have got a business case that we are in the middle of right now that is profitable.
7000 And all we are saying is that this Commission has chosen just to recognize that the market isn't as competitive at the consumer level as it should be. So what we are trying to do is find a vehicle to create more competition.
7001 Again, as I said this morning in the least intrusive manner, which is to give somebody an opportunity to buy something at a wholesale level, add on innovation, creativity and provide competitive services that aren't necessarily "me too" services as well.
7002 So I start from the premise that you are all competitive, the cable industry, the phone company industry are competitive on making money. And if you are and at the retail level you have got a price point, you back out those things you are not going to be incurring costs on and you add back those things that are going to be driven by this wholesale regime. Then you let the other guys take a shot at it and see whether that will create the competitive marketplace for Canadians.
7003 So somewhere in there retail-minus and cost-plus come together in a stage where hopefully the numbers work out. And I guess it's all of our jobs collectively to find that number or those numbers.
7004 MR. BIBIC: Well, I would say -- back to last Monday -- AVP is the vehicle. We have come forward.
7005 COMMISSIONER KATZ: But I didn't think your 17.8 cents, though -- and I try and work out and I'm not the engineer here, but the work that Chris Allen did this morning on the B.C. Broadband Association where he took a rate, a speed rate, and converted it into a usage rate and ended up suggesting that 11 cents per gigabyte is equivalent to $35 for 1 meg per second sustained.
7006 I'm looking at 17.8 cents here and my simple mind says 17.8 is more than 11 cents. Therefore, the $35 is going to be higher than that if I translate it and then I'm above retail.
7007 If I'm above retail it doesn't work, because then you are back to what you said earlier and that is the Commission is not going to allow a wholesale rate to be higher than a retail rate.
7008 MR. BIBIC: I am glad you agree with that because that's what is going to happen.
7009 COMMISSIONER KATZ: So you are saying you are providing services below costs today.
7010 MR. BIBIC: No. No. What we are saying is we are recovering those costs over the long term through tiered pricing. There is breakage inherent in our retail rates.
7011 In wholesale if you -- look, take AVP. We have come forward and we have said, well, we will live with cost-based. I mean, I'm not sure what more I'm supposed to say.
7012 We have stepped away from the market-based rates we were looking for, for AVP.
7013 We stepped away from retail-minus UBB.
7014 We have explained with empirical evidence and correlations that you asked for how volume correlates with -- Rogers showed how volume correlates with peaks at the nodes.
7015 We have shown how peak at the AHHSPI doesn't correlate with volume.
7016 We have come forward with that, but we have said we step away from pre and post-paid rates and we will take cost-based.
7017 Now, we will follow up. If you agree we will file our cost studies. You will approve a rate based on those costs and the Commission staff examination of it and we are done. It's cost-based.
7018 If we want to have a retail-minus then let's go back to UBB. But I don't think that will meet with much approval. We have already tried that one.
7019 COMMISSIONER KATZ: But if we end up with cost-based rates that are higher than retail we can only assume you are either selling below cost and that is either an interim period of time or there is something wrong with the marketplace.
7020 MR. DANIELS: Just to be clear, the model that we have filed and proposed for AVP underlying it we don't think results in a rate that is above retail.
7021 What we said is what would happen if we went to a different model is that you change behaviour and the costs, if you properly did the cost study, would actually result in costs that are above retail.
7022 COMMISSIONER KATZ: What does this 17.8 cents translate into a bandwidth rate?
7023 MR. DANIELS: It's 39.15 dollars using -- it's the exact -- if you go back and you do the model as to what underlines it, the exact same conversion ratio, because underlying both those is the exact same cost.
7024 COMMISSIONER KATZ: So it's 39 dollars and in there you are not recovering billing customer service, sales and marketing costs?
7025 MR. DANIELS: That's correct. We are recovering --
7026 COMMISSIONER KATZ: And how are you selling the service for it -- retail?
7027 MR. DANIELS: Right.
7028 COMMISSIONER KATZ: And how much are you selling that retail service for?
7029 MR. DANIELS: Hold it. It's $39.15 not per end-user. Let's be very clear it's per megabit per second. You pay that per megabit per second --
7030 COMMISSIONER KATZ: Right.
7031 MR. DANIELS: -- in the month.
7032 So like what that -- if you want to see a translation of that you can look at -- I don't want to take us through all of this but --
7033 COMMISSIONER KATZ: Well, you run into lanes per traffic and you split it by number of people that are going to be using it.
7034 MR. DANIELS: We did it in terms of answering Commissioner Morin's question about comparing MTS's pricing, and so on.
7035 So if you want to look at see what it looks like on a per-end-user basis using the assumptions of 100 end-users and a thousand so on, it's in that interrogatory -- sorry -- undertaking answer we filed today. So you can look at that.
7036 I'm not saying we agree with those assumptions but we just calculated. You will see the number there that you end up with is below the retail.
7037 COMMISSIONER KATZ: But I wouldn't mind seeing your -- if you don't agree with his assumptions, I would rather see your assumptions. I would like to see where you think the marketplace ends up based on your assumptions and relative to the retail marketplace that's out there whether there is a margin -- what the margin is left for the independent ISPs.
7038 You know what the retail market price is. You know what the rates are that you are charging. You know what these costs are. You know how to translate those costs as well.
7039 MR. DANIELS: So, if you are just saying can we take it and can we take it and put it on a piece of paper in terms of based on an assumption we can show you as compared to our retail rate. Of course the key thing is with AVP what the ISPs will do with it is they will come up with different pricing and packages.
7040 So that compared to our retail rate, yes, I can do that. I don't want to be making assumptions about what the ISP is going to do on their retail rate.
7041 MR. BIBIC: We will do that, Commissioner Katz, but recognize that when we show what that AVP rate with access compares to our retail rate, we shouldn't assume -- in fact, we do not want to assume that the ISPs will mirror our retail plan.
7042 COMMISSIONER KATZ: That's right.
7043 MR. BIBIC: They could maybe all offer the service for $20 more than us but they have the unlimited value.
7044 COMMISSIONER KATZ: That's right.
7045 MR. BIBIC: Unlimited pricing plan which adds value.
7046 COMMISSIONER KATZ: Yes.
7047 MR. BIBIC: Recognize that when we have a rate in the market of $39.95, that that's not all we are necessarily generating because we do have RPU associated with usage and recognize that there is --
7048 COMMISSIONER KATZ: Well, all I am saying is at a minimum --
7049 MR. BIBIC: -- associated with usage.
7050 COMMISSIONER KATZ: -- at a minimum they shouldn't be materially disadvantaged on a like-for-like basis. That's all.
7051 MR. BIBIC: And we are saying with AVP, we agree with you when it comes to AVP. We disagree with you when it comes to the other models because they will generate extra costs on our network.
7052 COMMISSIONER KATZ: Okay. No more questions.
7053 THE CHAIRPERSON: Tom?
7054 COMMISSIONER PENTEFOUNTAS: Yes. I would like to see those numbers as asked by Mr. Katz.
7055 That being said, you know, in your original submission you came up with a .195 rate on the pre-purchase and a .295 on the excess usage and today you are coming in at .178. Explain.
7056 MR. DANIELS: What we did is -- there has been an argument all week that we should have a cost-based rate. So the Commission approved last year in terms of what the mark-up is allowed on access and in the decision they laid last year in May they established for the first time for us, "We want you to have this set mark-up. It's confidential".
7057 And then in the decision that followed in September they said an extra 10 percent for FTTN, for our next generation service. So when we filed --
7058 COMMISSIONER PENTEFOUNTAS: What has changed between, I don't know, April and July or May and July. That's all old news.
7059 MR. DANIELS: But we are using now for the usage the same mark-up. When we filed 19-20, you know 19.5 cents and 29.5 cents we were acknowledging we had put in an even bigger mark-up. We have taken that down to match the mark-up mandated by the Commission. That's what we have done.
7060 So we have matched exactly what cable has said they were doing and what others have asked for, for a cost-based rate plus the Commission-approved mark-up.
7061 MR. BIBIC: Well, can I try? We did not -- the Commissioner's comments -- but we took the costs and added a higher mark-up.
7062 The 19.5 cents reflects a higher mark-up than what the Commission mandated for access. We thought that was appropriate in what we filed because we said an economic ITMP is meant to achieve a different purpose than access. It's justified to have a market-based mark-up.
7063 So 19.5 simply reflects a higher mark-up than 17 point. We backed away from it --
7064 COMMISSIONER PENTEFOUNTAS: Yeah, nothing has changed. Materially nothing has changed.
7065 MR. BIBIC: Yes. The cost -- the price is different. That's all.
7066 COMMISSIONER PENTEFOUNTAS: Okay. The costs haven't changed?
7067 MR. DANIELS: The costs haven't changed. Underlying it is the same cost study.
7068 COMMISSIONER PENTEFOUNTAS: Okay. That being said, I just want to -- and this refers to Bell and maybe it refers to CNOC more than Bell.
7069 I think it was -- I was under the impression last week that it would be pretty difficult -- we used the term gaming, but maximizing what you have purchased. That it would be pretty hard to come in close to whatever that 95 percentile was.
7070 I made the mistake, and perhaps that is the issue -- perhaps that is why CNOC prefers the 95 percent model in that there is a play there whereas with volume-based there is no play.
7071 And maybe CNOC can answer that question. Is that correct?
7072 MR. TACIT: That's a very big part of it. And in addition to that, even under a volume-based charge, for example, just to take a competing IPTV or over-the-top video service, for example, if you assume about two hours a day of usage over the course of a month we would have to pay $76.80 to Bell for that.
7073 So this is exactly the kind of situation whereas under the peak model we would pay for the peak traffic and everything underneath it and we could use it creatively to provide video, to provide Internet, whatever.
7074 COMMISSIONER PENTEFOUNTAS: But if you are successful then there are potential costs in that model that Bell would -- it would be impossible for them to recuperate.
7075 MR. TACIT: I disagree.
7076 COMMISSIONER PENTEFOUNTAS: You disagree?
7077 MR. TACIT: Yes.
7078 There are three issues here: What are the costs? How do they manage the network? How do we recover the costs?
7079 In my mind, the ILECs are trying to amalgamate all of these issues and it's creating a lot of confusion.
7080 They have costs for managing the network. We contribute to that whenever we contribute to network augmentation. The 95th percentile is a way that disciplines us to conserve that capacity. But at the same time, once we have paid for it, to use it and differentiate our services. And then, they have to manage their network.
7081 We are not saying we are not going to pay our share of the costs. If there's some flaw in the cost studies they have filed, they can update them or whatever. We are not suggesting they shouldn't recover their costs.
7082 COMMISSIONER PENTEFOUNTAS: You have to agree with me that, potentially, if we follow your model or the MTS model, that there are potential costs they would not be allowed to charge for if you played and successfully --
7083 MR. TACIT: I don't see it --
7084 COMMISSIONER PENTEFOUNTAS: I'm sorry?
7085 MR. TACIT: I don't see it at all. I can't conceive of what costs we wouldn't be recovering.
7086 If this Commission says Bell's costs are X, and then backs a rate out that's meant based on the assumed demand for peak capacity in the marketplace to recover those costs, then they are going to recover those costs.
7087 I mean any system is subject for forecast errors. If that's your concern, then that concern exists under any model, if we are talking about forecasts.
7088 COMMISSIONER PENTEFOUNTAS: Well, no, but they would have to bill you differently. If you can achieve something close to 95 peak, then their cost model changes completely. Therein lies the problem, I think.
7089 MR. TACIT: They have to know how to model costs under that scenario. Yes, it's a shift. Obviously, we are rate restructuring, so, yes, they need to make a new set of assumptions in order to recover their costs. Fine, let them make a new set of assumptions to recover their costs.
7090 COMMISSIONER PENTEFOUNTAS: Okay.
7091 THE CHAIRPERSON: Let's move on. We have heard this before.
7093 COMMISSIONER MOLNAR: My first question actually just follows this conversation.
7094 This went on earlier and Mr. Tacit suggested that if there are significant changes in the traffic patterns caused by putting in a billing model such as a capacity-based billing model that we could reviews costs in the future.
7095 You did not respond to that, but isn't that in large part some of your concern here?
7096 You said your revenues will be the same, your costs won't be the same because you expect traffic patterns to change, as I understand it, the ISP's traffic pattern's to change, and that you will incur larger costs. It's you expect it to happen. I don't recall on this record any evidence that will occur, but there's an expectation it could occur.
7097 Can we address this simply by making a commitment to review these costing studies in a certain number of years?
7098 MR. BIBIC: Absolutely not, Commissioner Molnar.
7099 Again, you are jumping the second phase, which the vice-chairman explored, about what the impacts are going to be and the inability to recover those costs if you were, the Commission, to impose or adopt or approve 95th percentile or capacity.
7100 I start one step earlier. So does cable. It's measuring the wrong thing.
7101 I mean you asked for correlations. We have given you the correlations, or the lack thereof.
7102 COMMISSIONER MOLNAR: Well, Mr. Bibic, that would give me my second question.
7103 You gave us the correlations, you said. Could you tell me where on the record you correlated total volume to investment?
7104 MR. BIBIC: Total volume to investment. Investment where? At which point?
7105 COMMISSIONER MOLNAR: Well, did you correlate total volume to investment at any point?
7106 MR. BIBIC: In other words, correlate the total volume with where we have to go in and augment a link? Is that what you mean?
7107 COMMISSIONER MOLNAR: Well, if you have done it at the link or if you have done it on an aggregated basis, have you done it somewhere? Is it on the record?
7108 MR. DANIELS: I think that's what cable showed you earlier today.
7109 You see, the difference between their correlation --
7110 COMMISSIONER MOLNAR: But I asked if you did that --
7111 MR. DANIELS: Have we done that?
7112 COMMISSIONER MOLNAR: -- not if cable did that. Did you do that?
7113 MR. DANIELS: Well, we didn't file the document earlier today, but we agree with that assessment if we looked at it in terms of the various links. I think they showed you the correlation between total volume, because you said there's no evidence, and I think that's what we actually did.
7114 Now we did file our total growth and we filed our investment figures, and you can see that there is a relationship there. From the very aggregate, yes, we continue to invest, and we are experiencing the growth, on the high level there. But in terms of looking at is there a correlation on a particular link, I think cable showed you that today. And what we are showing --
7115 THE CHAIRPERSON: Yes, so that begs the question: are you willing to file the same thing as cable did? That's what my colleague is asking.
7116 MR. BIBIC: Yes.
7117 And then to go back, if I may, Commissioner Molnar --
7118 COMMISSIONER MOLNAR: Well, I would like to just --
7119 MR. BIBIC: Well, you asked me --
7120 COMMISSIONER MOLNAR: I will let you go back in a second --
7121 MR. BIBIC: Okay. All right.
7122 COMMISSIONER MOLNAR: -- but I was interested in whether or not you did file a correlation between volume and investment.
7123 I would also be interested if you could do the same, then, between peak usage and investment.
7124 If you can recreate the cable for DSLAMs -- I assume that's what you are going to do is recreate this for DSLAMs, because they did it at the CMTS -- and you are going to correlate total volume to the -- what did they do? They did the total volume to peak node utilization, and you are going to do the same for a DSLAM. Can you also do that on peak usage?
7125 MR. BIBIC: They will confer.
7126 MR. BIBIC: We know what you are driving at and we are going to file whatever we can to get to the points that you are driving at. It's just that on the spot they do not know exactly what they will be able to pull.
7127 But questions were asked last week. We managed to pull the correlations. We will do our best and we will file what we have.
7128 COMMISSIONER MOLNAR: Fair enough.
7129 As you said, you know what I am driving at. I am driving at the point that, while there are statements that total volume correlates to investments, I don't know that I have seen the evidence that supports that.
7130 You have said it at the disaggregated right down to distribution of the DSLAMs. And that's your concern: the aggregate doesn't reflect what occurs deeper in your network. So if you do have that, I would appreciate that.
7131 MR. BIBIC: We understand what you are looking for.
7132 But could I go back to where you started, which is that our case has been the 95th percentile or the other models will cause more costs in the aggregation network, and there's no evidence of that necessarily, so is one solution to review costs down the road. And you did say that you see no evidence.
7133 Well, I'm not sure. We spent the last eight days and we have Primus saying that of course they are going to use what they pay for the most efficiently.
7134 Mr. Tacit just finished answering the vice-chair basically with an answer I would have given, which is: of course, we are going to do this, and there's nothing wrong with it. I agree.
7135 You have Mr. Coyne last week saying, yes, there can be congestion in the aggregation network at times other than the peak that I'm causing, and that may be due to me, and that's not reflected in the current rate -- I mean the quote's right in my rebuttal argument -- and that's for the network provider to manage.
7136 We have what we see in the U.K., where things like creative packages like Midnight Madness sales do occur. I mean these are all points of evidence. It's going to cause congestion in the aggregation network. It's going to cause greater costs.
7137 So as I sit here, I say: why would we review that five years from now? If you were to adopt one of these models, as imperfect as it is, then you have got to reflect those assumptions today. Not tomorrow, not in five years, today.
7138 We don't think you should go there because we think AVP's better --
7139 COMMISSIONER MOLNAR: Mr. Bibic, I'm not saying that people haven't acknowledged that under a capacity model there would be incentives to use that capacity efficiently and effectively, and to use peak, or as close as possible. But what I'm saying is in Figure 4, you have shown what could happen, where it will change the traffic pattern significantly and cause you to incur different costs deep within your distribution network because the traffic mix changes.
7140 I'm just saying we haven't seen the traffic mix change, that they might raise their traffic. And I'm saying that it would change and cause you to augment your network.
7141 I mean every time I hear about the Midnight Madness, I kind of think: So what? I'm not sure I have seen anything on the record that would cause me to be concerned with Midnight Madness.
7142 If it doesn't cause you additional costs, so what?
7143 So if there's something I'm missing, either because it's filed and I haven't seen it, let me know. Because if it's going to cause you additional costs, and it's not on the record today -- that was my question -- when there is evidence that there are changes, can we address this in the future? Is there some assurance in the future, through recosting, you know, if things happen in a way unanticipated in this proceeding, that we can --
7144 MR. BIBIC: Well, Commissioner Molnar, you have our position.
7145 COMMISSIONER MOLNAR: Okay.
7146 I want to ask you about the MTS and Primus position because I am a bit confused as to why you view it to be so much worse than the 95th percentile.
7147 And I do want to put the issue of the flat rate access aside, just talking about their proposed use of a single rate at the interface.
7148 Could you just clarify again why that is of significant concern?
7149 MR. DANIELS: I think, first and foremost, in terms of doing it, we start from the premise that, in terms of putting a fair model before the Commission, it's pay for what you use. We believe AVP is the best way to do that. There's some who argued about 95th percentile, but the MTS model underlying it is pay for what you order, which you may or may not use.
7150 We have heard other people raise all sorts of issues that they are going to have, and to be completely blunt about it, what's going to happen, we think, is that, one, we have that same concern about increasing the costs -- I think you understand that and I'm not going to repeat that in terms of what the impact will be -- but on top of it we have a feeling that the Commission's going to look at the number and say: Well, if you're actually provisioning as if there's committed bandwidth all the way to our COs, which is what the MTS model does, it's going to result in pretty high costs that aren't going to be acceptable to the Commission, which I think Mr. Bibic -- you know, Mirko covered in the opening statement today.
7151 I can just see the type of issues -- and it's been raised -- oh, what about redundancy? Okay, so let's just reduce all of the accesses in half or so on.
7152 You are either going to have to start manipulating it, I believe, in order to create something that's acceptable to the ISPs or you are going to leave in place rates like MTS has today for its HSPI for Legacy, 1 gig, $68,500 per month -- per month -- compared to our $1,600, and you are going to have that in place today for Legacy and you are going to end up with the results of no customers.
7153 I don't know why -- well, I'm pretty sure I know why no ISPs -- there's not significant ISPs, I shouldn't say no ISPs in Manitoba -- and it has something to do with that rate, I suspect strongly. And so I think CNOC et al can agree on that point.
7154 So you look at it and you come back and say: Okay, well, what about different rates or a lower rate and so on and run the costs model this way, make this assumption, and so on and so forth, let's get that down. That's fundamentally our concern.
7155 Conceptually, if you said to us: Oh, sure, you're going to be paid for more bandwidth than is actually going to be provided and make it that it doesn't work for the ISP, why should you care Bell? Because I think, ultimately, we know that we have to have a model that works.
7156 It has got to work for us, but it has also got to work for the ISPs, and we are afraid that model will lead to one that, ultimately, because it won't work for the ISPs, you will have to make assumptions, you will have to make changes, and ultimately to the detriment of Bell.
7157 COMMISSIONER MOLNAR: Thank you. Those are my questions.
7158 THE CHAIRPERSON: Marc?
7159 COMMISSIONER PATRONE: Thank you, Mr. Chairman.
7160 Good afternoon.
7161 Along this same vein, under the MTS/Primus model, is there anything that would prevent you from negotiating carrier-specific agreements with the wholesalers that essentially offer to buy back, in the form of either rebates or credits, unused capacity?
7162 As I see it, their complaint is that, under the MTS model, they don't want to end up with a lot of capacity that they are not going to end up using. Your complaint is that they are going to end up with capacity which they are going to want to fill up through Midnight Madness deals.
7163 But they may turn around and say: Well, rather than offer a Midnight Madness deal, why not simply offer it back to you and treat broadband as a kind of commodity that you could negotiate the buy and sell of, if you know what I mean. That would bring you back, conceivably, to a volume-based level. It would be kind of a round-about way in which you would reach that point.
7164 Is there anything preventing those types of agreements?
7165 MR. BIBIC: I think we haven't done a very good job on either side reaching an agreement on these issues to date. In theory, could we do it? Yes, I suppose we could. I think evidence shows that we haven't been able to reach commercial and off-tariff arrangements on the central issues to now, so I have some doubts about that.
7166 You do properly characterize our concerns with the MTS model and I think you do properly characterize the ISPs' concerns with it. In fact, you know, Mr. Cohen is not in favour. We heard BCBA say they are not in favour. The Primus model, they say, is their second choice; the 95th percentile is their first choice.
7167 A completely separate issue that we have is that -- and this I covered this morning -- the MTS model actually undermines the speed-matching decision because it gives the ISP access to our network, in the MTS case 32 megabits per second access, and they are not constrained by the speeds that we offer in the marketplace. I find it very, very odd that, and I mentioned this morning, we can't offer anything at retail, in terms of the speed tier, unless we offer at wholesale, but now wholesale ISPs could take our network and do what we are not permitted to do. So that's a separate concern.
7168 But to your specific question, Mr. Patrone, I guess I struggle with it because I'm not sure we would be able to get there.
7169 And it sounds to me like we would be back to the 95th. If an ISP buys the full capacity, but then doesn't use it and sells it back to us, presumably we are not going to take back to a level lower than their 95th percentile peak, so we end up back at 95th --
7170 COMMISSIONER PATRONE: Well, if you were buying unused capacity, wouldn't it bring you back to volume-based? They would basically be selling you back or in the form of either a credit going forward or a rebate, capacity that, in their view, is unused.
7171 MR. DANIELS: I think as part of those negotiations you are going to end up having the problems of are we buying back gigabits, are we buying back the 95th percentile peak disputes, back before the Commission, will be right around here.
7172 COMMISSIONER PATRONE: Presumably, though, Mr. Daniels, that could be subject to negotiation, couldn't it?
7173 MR. DANIELS: Listen, anything could be subject to negotiation except one thing: we certainly never would negotiate about what a wholesale customer could do in terms of -- I would never want to have a negotiation that said: Oh, don't offer Midnight Madness and we'll give you this discount.
7174 Like, that will never happen. That's not allowed and illegal and we won't have that.
7175 COMMISSIONER PATRONE: What you are suggesting is that the Midnight Madness stuff, at least as I have understood it, would be their way of dealing with their unused capacity. When you offer that kind of a deal, presumably, it's at a loss or you are not making a whole lot of money by offering a Midnight Madness deal. Maybe it's their incentive. They would be better off selling it back to you in the form of a credit or a rebate going forward than they would basically dumping it in the form of commercial offerings that don't bring in any profit for them.
7176 MR. DANIELS: Well, I guess the problem we would get into is just trying to figure out how we quantify where would actually -- the only thing we would want to buy back in that would be the actual cost savings, which would have to be figured out, oh, if you use the network less, which is back to how do you determine what would actually happen.
7177 I don't rule out anything as possible, in terms of negotiation, but on a shared network we would have to go back to try and identify we would actually be saving by them not sending the traffic, which is back to the same issue we had before. We would rather just get there directly through the AVP.
7178 COMMISSIONER PATRONE: Okay.
7179 Well, before I hand it back to the Chairman, I mean it does strike me that there would be incentive on both sides to deal in a case like that.
7180 THE CHAIRPERSON: Okay, I think we have had a full discussion. You made me break my golden rule of always finishing by 4:30. It's 4:56 now.
7181 That's it for today. I'm finished, and I think so are you.
7182 THE CHAIRPERSON: We will start tomorrow at 9 o'clock.
7183 Thank you.
--- Whereupon the hearing adjourned at 1656, to resume on Tuesday, July 19, 2011 at 0900