ARCHIVED -  Transcript of Proceeding

This page has been archived on the Web

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please contact us to request a format other than those available.

Providing Content in Canada's Official Languages

Please note that the Official Languages Act requires that government publications be available in both official languages.

In order to meet some of the requirements under this Act, the Commission's transcripts will therefore be bilingual as to their covers, the listing of CRTC members and staff attending the hearings, and the table of contents.

However, the aforementioned publication is the recorded verbatim transcript and, as such, is transcribed in either of the official languages, depending on the language spoken by the participant at the hearing.





Review of the Internet traffic management practices of Internet service providers


Conference Centre

Outaouais Room

140 Promenade du Portage

Gatineau, Quebec

July 14, 2009


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


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 the Internet traffic management practices of Internet service providers


Konrad von Finckenstein   Chairperson

Len Katz   Commissioner

Suzanne Lamarre   Commissioner

Candice Molnar   Commissioner

Timothy Denton   Commissioner


Sylvie Bouffard   Secretary

Regan Morris   Legal Counsel /

Chris Seidl   Hearing Managers

Stephan Meyer


Conference Centre

Outaouais Room

140 Promenade du Portage

Gatineau, Quebec

July 14, 2009

- iv -



Bell Aliant Communications Partnership and Bell Canada   1049 / 5971

   Gatineau, Quebec

--- Upon resuming on Tuesday, July 14, 2009 at 0830

5965   LE PRÉSIDENT: Bon. Commençons, Madame la Secrétaire.

5966   LA SECRÉTAIRE: Merci, Monsieur le Président.

5967   Before we begin, I would like to note for the record that due to a power shutdown of the CRTC complex and access to our website, the date for final replies has been extended to end of day Tuesday, 28 July 2009.

5968   We will now proceed with the presentation by Bell Aliant Communications Partnership and Bell Canada.

5969   Appearing for Bell Aliant Communications Partnership and Bell Canada is Mr. Jonathan Daniels.

5970   Please introduce your colleagues and proceed with your 15 minute presentation.


5971   MR. DANIELS: Thank you, Madam Secretary.

5972   Mr. Chairman and Commissioners, I am Jonathan Daniels, Vice-President, Regulatory Law for Bell Canada. I am pleased to introduce our panel appearing on behalf of Bell Aliant and Bell Canada, which collectively I will refer to as "the Companies".

5973   Joining me today, to my immediate left is Carl Condon, Vice President if Network Technology and Planning; to his left is Denis Henry, Vice President of Regulatory and Government Affairs and Chief of Privacy for Bell Aliant.

5974   To my immediate right is Peter Vandenengel, Director of Bell Internet Product Management; and to his right is Suzanne Morin, Assistant General Counsel, Regulatory Law and Policy and Bell Privacy Ombudsman for Bell Canada.

5975   Fundamentally, at the heart of this hearing is a dispute about innovation. Some have argued that application specific traffic shaping harms innovation. Without one iota of proof, these parties simply assert that the threat posed by application specific shaping is so severe, that it must be stopped.

5976   Underlying this position is a belief that application innovation requires dumb networks and activities which add intelligence to the network have the potential to threaten the ability for innovation to occur at the application layer.

5977   However, we submit to you that the development of innovative products and services requires innovation at both the application and network level.

5978   We cannot ignore the history of innovative networks and the potential societal benefits from future innovation at the network level. Where some see the need to impose regulatory rules to protect application innovation, we see their demands as seeking regulatory rules that would limit, or even ban, innovation at the network level. Quite simply, imposing ex ante rules that limit or ban innovation at any layer is by definition a greater threat to innovation than having no ex ante rules at all.

5979   With respect to the question of the open Internet, we do not believe that there really is any dispute. No one is arguing for the right of ISPs to make editorial decisions with regard to which content should and which content should not be blocked. We can all agree on that point.

5980   At the outset, I note from the discussion that has occurred over the past week that the congestion problem and the role P2P plays in causing that problem has been well canvassed so I will not spend any more time on that in my opening statement.

5981   To manage capacity on its wireline network, Bell Canada has adopted a three-pronged approach. It is important to note that each of these activities on its own is not sufficient to address network congestion. Effectively addressing network congestion requires using a combination of all three of these activities.

5982   Our first prong -- and I will briefly touch on all three -- is capital spending.

5983   Bell Canada continues to invest in its network. We have spent close to half a billion dollars in each of the last two years to expand and upgrade the infrastructure of our DSL network, and we plan to continue to invest.

5984   Unlimited capital spending is not a luxury the Companies' enjoy. The economic and market realities of high-speed Internet access clearly show that a pure capital spend solution is not a sustainable model. It is also far too simplistic to assume that we can build our way out of the problem using revenue from customers who some argue are willing to accept substantial price increases in order to use a network that is not subject to traffic shaping.

5985   Turning, then, to the second prong, pricing and new business models.

5986   Some parties have argued that prices can be used to solve problems of congestion. We agree that prices provide the appropriate incentives for optimal network use and expansion. However, in today's market realities prices are not granular enough to eliminate congestion during the peak hours. Prices motivate users to reduce consumption over the month and deal with average usage.

5987   Our problem is during the peak and so in terms of overall usage during the month having an overall cap in that regard doesn't directly address our problem of congestion during the peak period. That's why I say, while pricing helps, it is also not the sole solution.

5988   Turning, then, to the third Prong, bandwidth management, we firmly believe that our traffic shaping practices are acceptable. They are designed to improve the overall user experience for the vast majority of users. The practices we implement are:

5989   One, narrowly focused and limited to non-time sensitive traffic;

5990   Two, confined to a defined period of time; and

5991   Three, do not block access to content.

5992   However, the Companies do not want to suggest that current practices will never adapt to changing circumstances. The Companies are committed to continually monitoring and evaluating the effectiveness of our practices in managing the impacts of real traffic on our networks.

5993   Several parties and Commissioners have asked about potential privacy concerns regarding the implementation of ITMPs and, in particular, the use of DPI technology for traffic shaping.

5994   We can confirm that DPI deployed in our network is being used only for the purposes of traffic management. Our DPI does not, and cannot, inspect the user content of communications. Data collected is used at the aggregate level to better understand the nature of traffic on our network and identify trends such as increases in the use of particular applications. It is not being used for marketing purposes.

5995   Now, I would like to spend a minute or two on the wholesale issues that we have heard in this proceeding.

5996   We have found it perplexing to hear some ISPs state that they should have the right to manage their own networks using the same tools as the Companies, but in the same breath argue that strict rules are required regarding how we manage our network when they buy GAS service from us.

5997   As we listened over the last week or so there are two assertions that must be corrected.

5998   First, contrary to what many have told you last week, retail users and wholesale GAS users share the same network. Wholesale GAS and retail Internet traffic travels across the same network elements and links and is commingled in the aggregation portion of our network. At that point, traffic is not segregated either physically or it is also not segregated virtually.

5999   This aggregation portion of our network is the exact area where we have our largest congestion problem. As a result, and to put it straightforward, wholesale traffic impacts retail traffic and vice versa.

6000   Second, wholesale ISPs have attempted to dismiss their impact on our retail network by pointing to their overall market share. The reality is that wholesale traffic, at least on our network, is significant. In the month of May 2009, the most recent month for which data is readily available, wholesale GAS traffic accounted for 31 percent of our total GAS and retail traffic. That is over twice as high as the percentage of end users wholesale represents on our network. Further, GAS traffic accounted for 36 percent of our total P2P traffic.

6001   The Commission must also not forget that wholesale customers do have options. If they lease unbundled loops, they can choose not to shape, or implement their own ITMPs, as Primus does, when they lease our unbundled loops. That is their reference to their on-network. When they talk their on-net before you they were referring to them leasing our unbundled loops and being able to avoid our traffic shaping policies and implement their own ITMPs.

6002   Mr. Chairman, one party noted that Bell had implemented traffic shaping only on retail before extending it to wholesale and used this fact to conclude that it was unnecessary for Bell to shape wholesale traffic.

6003   You rightly stated that another possible conclusion was that we implemented shaping on retail traffic and found that it was not sufficient so we decided to extend it to wholesale.

6004   However, the reality is that the problem has always been both retail and wholesale. We decided to focus on the retail market first in order to ensure that we got the solution right before we extended it to wholesale.

6005   In the face of complaints, the Commission will need to determine whether an ITMP which discriminates, constitutes, in the words of subsection 27(2), an "undue or unreasonable preference". The statute requires a reasonableness standard.

6006   Now, some want you to adopt an "Oakes-type" test, including the least intrusive standard. Inherent in the nature of any test that requires the determination of "least intrusive" is the notion of trying to determine a single "best" method -- at least for a particular ISP -- rather than considering whether any particular practice is reasonable.

6007   And who determines the best method under Oakes? Well, ultimately in the face of complaint I think that would have to be you, the CRTC. With all due respect, the appropriate test for ITMPs is not one which has the regulator substituting its opinion for that of market participants.

6008   The Oakes Test is a constitutional test. It is fundamentally about limiting government intervention into private affairs. We agree with ARCH when they stated:

"... this is not the time to refashion the interpretation of section 27 ... If the ... Oakes Test is used to interpret ... [section] 27, you will undo decades of clear and effective section 27 jurisprudence with respect to discrimination ..."

6009   In short, we submit that the appropriate test for reviewing an ITMP, and the one required by the statute, is a reasonableness test, not a least intrusive test.

6010   Now, we understand the Commission's desire to consider the broader implications of these practices for both retail and wholesale services in order to establish guidelines surrounding acceptable ITMPs. To that end, the Companies have developed guidelines for your consideration for the use of ITMPs in an effort to balance:

6011   (a) the desire of Canadians to access the Internet content of their choice, to use and create applications, and to use the Internet; and

6012   (b) the legitimate interests of ISPs to manage their networks, including capital spending, effectively and efficiently.

6013   These Guidelines would apply to all ISPs regardless of size, legal status and underlying technology. These Guidelines would apply to all ITMPs as set out in Telecom Public Notice.

6014   For greater certainty, they do not cover ISP business models designed to monetize networks. such as setting different service tiers or content arrangements, and they do not cover traffic that is carried on an ISP's network but not on the public Internet such as licensed IPTV service, managed VoIP, or dedicated business services.

6015   We propose the following three guidelines:

6016   One: ISPs should make reasonable efforts to limit the negative impacts of ITMPs on users, services, protocols or applications.

6017   Implicit in the determination of what is "reasonable" is the recognition that different networks face different problems and therefore ISPs require the flexibility to reflect those distinctions in adopting ITMPs.

6018   We have four examples here:

6019   First, ISPs may take into account real-time versus non-real-time nature of applications/

6020   Second, ISPs may take into account differences that may exist between wireline and wireless networks and even within wireline networks, differences between cable, copper and fibre networks.

6021   Third, ISPs may not implement ITMPs that give an undue preference to retail traffic over their wholesale traffic that share the same network.

6022   Fourth, ISPs may not implement ITMPs that block content based on its source, ownership or destination unless such activity is necessary to protect the integrity of the network. By that reference there we are thinking of blocking spam or malware.

6023   Two, our second, is transparency.

6024   ISPs should disclose to affected retail and wholesale customers the general nature of implemented ITMPs in a manner that does not compromise the security of networks and commercially sensitive information. For example, ISPs should disclose the kind of traffic affected, the times of day impacted and any privacy implications.

6025   ISPs should implement ITMPs in a manner consistent with applicable privacy laws. For example, data used for the purpose of ITMPs should not be used for any other purpose without customer consent.

6026   We believe that these guidelines address the six issues raised by the Commission, namely: acceptability of ITMPs, disclosure, privacy, wholesale and special treatment for wireless.

6027   As for the sixth item, section 36, it is our view that it is rare that section 36 issues will arise with respect to ITMPs. To the extent that they do arise, it is possible to address them under the first guideline which provides, as an example, that ISPs should not implement ITMPs that block content based on its source, ownership or destination unless such activity is necessary to protect the integrity of the network.

6028   Mr. Chairman and Commissioners, we have argued for a reasonableness test. We believe that our guidelines and the examples of reasonable ITMPs that form a key component of them -- with them, ISPs will have enough clarity under which to make their ITMP decisions.

6029   In short, these guidelines give substance to the notion of reasonableness in the context of ITMPs. At the same time, these guidelines are flexible enough not to restrict network innovation while protecting application innovation.

6030   By using a reasonableness standard, they reflect the dynamic and unpredictable nature of the Internet.

6031   I end where I began, imposing rules that limit or ban innovation at any layer is by definition a greater threat to innovation than having no rules at all. Our guidelines will ensure that innovation continues to occur at all levels of the Internet.

6032   For your ease of reference, on the next page following to our opening statement I have attached the guidelines in one page so that you can see what we are proposing.

6033   This concludes our remarks and we welcome your questions.

6034   THE CHAIRPERSON: Well, thank you very much.

6035   There is an advantage to being last, you have heard everybody else and you have an idea of what the Commission wants and so you can answer it. So you provided a set of guidelines as I asked Mr. Engelhart from Rogers to do.

6036   The disadvantage of course is you have an informed panel and so you are going to get a lot of scrutiny.

6037   So tell me, P2P which seems to be the flashpoint of all of it, we started out with hearing from some people saying that you actually cannot detect it, you only detect the most visible ones, that the DPI or whatever technology, there is a lot of P2P traffic which is encrypted and therefore you can't detect it so it passes through.

6038   What is your view of this situation? I mean you obviously are doing P2P monitoring and throttling. Do you have any evidence that you actually do catch a percentage of P2P traffic and what is it?

6039   MR. DANIELS: I am going to hand that over to my colleague, Mr. Condon, in a minute.

6040   I think the short answer is, the traffic shaping that we do is sufficiently effective to achieve the objectives, in terms of our overall user experience for all of our users on our network. We wouldn't be doing it if we didn't think it was sufficient to do that.

6041   The suggestion is that we may not be capturing all of the P2P out there, there may be some that we are not, and I am going to let Mr. Condon address that, but the first thing I can assure you of is that we believe our practices are acceptable, because we are doing it to improve the overall user experience.

6042   MR. CONDON: Commissioner, yes, I think that we do have a very accurate ability to recognize P2P. Evidence of that is that, overall, 27 percent of our traffic is peer-to-peer throughout the day, but at peak that number reduces to 14 percent as a result of our shaping of that traffic.

6043   So we can recognize it, yes.

6044   THE CHAIRPERSON: That is the P2P traffic that you identify.

6045   MR. CONDON: Yes.

6046   THE CHAIRPERSON: Do you ever do any sort of ex post facto analysis of the traffic that went through, whether there has been any P2P that's encrypted that you didn't pick up?

6047   MR. CONDON: There is some, Commissioner. We know that, but our position is that we prefer not to create false positives. That is just one of the implications of that.

6048   THE CHAIRPERSON: And I presume, like the others before you, when in doubt you let things through?

6049   MR. CONDON: Yes.

6050   THE CHAIRPERSON: You are the first one who has answered the question that Mr. Katz has asked everybody else, whether the P2P traffic on your wholesale customers is equivalent to or is roughly proportional to the total value.

6051   In paragraph 16 you suggest that 31 percent of your total traffic is GAS, and of that 31 percent, 36 percent of your GAS traffic is P2P.

6052   I gather from that -- and tell me if I am wrong -- that the amount of P2P traffic carried by your wholesale customer is no different than that carried by your retail customer, it's roughly proportional.

6053   MR. DANIELS: There is a slight difference --

6054   THE CHAIRPERSON: Yes, I'm sure.

6055   MR. DANIELS: -- but in terms of overall, it is roughly the same. It is 36 versus 31, so --

6056   THE CHAIRPERSON: I see.

6057   Now, we have heard from various ISPs, and they seem to go all over the place. Videotron says: We only use commercial pricing incentives, or peer discipline.

6058   We have heard from Shaw, who said: We actually monitor, and if it goes beyond a certain dangerous threshold, we start throttling, and then we take the throttling off as soon as the P2P traffic goes down.

6059   You are the ones who throttle during peak hours, and then we have Cogeco, who says: We throttle 24/7.

6060   What are we to make of it? Why are different methods being used?

6061   I go from the assumption that nobody throttles unless they have to in order to protect their network.

6062   Why is there such a variance of approaches?

6063   MR. DANIELS: I am going to take the first stab at that. I think the short answer is that each looks at it from their unique perspective of what their network -- what works for a cable company, and even between cable companies, which I can't speak to because we don't have expertise in that area, is very different.

6064   We looked at solutions that would address the issue in our network, based on the problems that we were facing, and the solutions that we saw that would address it in a manner which was to the benefit of our users.

6065   I think that every network can make their own determinations and face their own problems, which are unique, and that is one of our concerns about this notion of having the Commission dictate, ultimately, what some parties want, saying: This is the solution you should impose.

6066   We are not aware of any unreasonable practice in the Canadian market today, so that is the emphasis behind the policy piece.

6067   Carl, is there anything you want to add?

6068   MR. CONDON: No, I don't think that I have anything to add to that.

6069   THE CHAIRPERSON: Tell me why you cannot use the Shaw approach. It struck me as a very reasonable one, you do the measuring of your traffic, you identify P2P, you have a danger threshold where you set it, presumably, and there are also the technical considerations, and as you approach it, you start throttling and the traffic falls.

6070   Other than, as you do, doing it at peak hours, wouldn't that be the logical --

6071   You are advocating a reasonable approach. I would have thought that reasonable meant you only do it when you have to do it, rather than having a blanket across peak hours.

6072   MR. CONDON: I think there is a pretty big technical difference between the cable networks, Commissioner, and the telephone networks. Cable, as I understand it, is measuring at the CMTS. We do not have a CMTS in our network, and we don't have --

6073   THE CHAIRPERSON: You will have to translate that for me.

6074   MR. CONDON: Cable Modem Termination System, which is often known as the head end, where the cable modems are controlled.

6075   The equivalent of that in our network is the DSLAM, the Dual Subscriber Line Access Module.

6076   We don't face the same sorts of congestion. We don't have the same sorts of metrics.

6077   On top of that, our network is highly meshed, and congestion, as we said, occurs in the Aggregation Network, and we have about 1,500 links in that network. The DPIs sit above a number of those links. Some of those could be congested, while, probably, others are not.

6078   You cannot manage on an individual link basis on a network like that.

6079   THE CHAIRPERSON: I am trying to dissect from your answer whether this is an economic problem or a technological problem.

6080   MR. CONDON: Technological.

6081   THE CHAIRPERSON: So even if you wanted to, if you had the money, you couldn't technologically do it.

6082   MR. CONDON: That's correct.

6083   THE CHAIRPERSON: Okay. Now, Mr. Daniels, you said in your submission, and also in today's presentation, that economic incentives or disciplines do work; however, in today's market, prices are not granular enough to eliminate congestion during peak hours.

6084   What stops you from making pricing more granular to have that effect?

6085   MR. DANIELS: I am going to hand that over to my colleague from our Product Department.

6086   MR. VANDENENGEL: Thank you, Commissioner.

6087   As we pointed out, the pricing mechanism that we have today is not granular enough to affect our peak period issue, so in order to have a pricing mechanism work during peak periods, we would have to have pricing built on peak period usage.

6088   I think that Sandvine, on Monday, pointed out some of the problems with that. One, it would require an education of the base to understand what usage is about, what different applications pull different amounts of usage.

6089   It would also require pricing to happen at the time when they wanted to use it the most. So either they would be using it less when they wanted to use it, or they would actually be incurring price increases.

6090   And the third aspect is, I think, to highlight it, and we reflect that, as well as the upgrade of the billing system. We would have to inject a lot of money into our billing systems to get that capability in place.

6091   THE CHAIRPERSON: What is the difference between internet use and wireless phone use?

6092   Wireless phone use -- all of the various plans give you different price rates, whether you make long distance calls, or you use it after hours, or during peak hours, et cetera. The public seems to be very -- and if you don't want it, you can buy global plans, because there is various pricing.

6093   You have, very well, educated consumers on how to use their wireless phones, why can't you do the same thing here?

6094   MR. VANDENENGEL: I think it goes back to education at the base level. Wireless started that aspect many, many years ago. They took that route on their network years ago, and they started educating the base.

6095   I am not saying that it's not possible. The three items I laid out are what wireless did, and we would have to do the same thing to move in that direction.

6096   The issue with that is that, in the end, it is going to lead, in my opinion, to price increases, and the vast majority of customers do not want the price to go up in this type of market.

6097   THE CHAIRPERSON: Or is it that you are afraid to be the leader and be at a competitive disadvantage, while others try to lure some of your customers before they are forced to do the same thing?

6098   MR. DANIELS: Part of that answer we have submitted to you directly, in confidence.

6099   If I could get you to turn, if you have our submission --

6100   THE CHAIRPERSON: I read your submission.

6101   You say on the record whatever you feel free to say.

6102   MR. DANIELS: I think what we are comfortable saying on the public record about that is that the market is not there yet, and should the market move there one day, it is possible in the future, but there can't be an overnight change to the market in that regard.

6103   The other aspect, I would say, in terms of what is different between wireless and internet usage, is that customers understand when they are using their wireless -- when they make voice calls, they understand how much bandwidth it uses. It's pretty clear how the pricing mechanisms work.

6104   But when you are using the internet, some applications are much more bandwidth intensive than others, and customers don't fully understand that yet, and that's part of it.

6105   You see some of the problems that have happened in the market -- major stories -- in terms of customers understanding it when they use wireless, so I am not sure that that model is going to be able to continue to thrive in the near future in the wireless market either, when it comes to data broadband. We will have to see.

6106   Customers -- it requires a level of understanding of their bandwidth, what different applications use, that the average person doesn't understand.

6107   And I come back to the notion of what is the problem that we are really trying to address here, because the vast majority of users are happy with the nature of how the internet operates.

6108   And when I say that, I don't just mean on our network. We submitted to you the Harris/Decima study, which stated that 87 percent of users are happy with their internet today.

6109   So is this problem, which we believe -- and there are a number of reasons that we focus on P2P, but we think that it improves the overall user experience, and I think you have heard that time and time again from various different ISPs.

6110   The trade-off of going to a completely peer usage-based billing practice on the time of day, and having to understand how much bandwidth each application takes, and so on, I think, would add a layer of complexity which would undermine the use of the internet rather than promote it.

6111   THE CHAIRPERSON: That is a long way of saying that you think wireless is going to go the way of internet pricing, rather than the other way around.

6112   MR. DANIELS: I am not sure what is going to happen, I am just saying that I am not convinced that it won't.

6113   THE CHAIRPERSON: You heard Rogers yesterday testify here, and they basically said that economic incentives or disciplines only go so far, because they only deal with the downloading issue, they don't deal with the uploading issue, and that is really where the problem is.

6114   And even if we did introduce such pricing, we would then have to educate people on the upstream, and what it does, and how it works, and whether to turn it off, et cetera.

6115   What is your view? Do you agree with Rogers that introducing economic disciplines or incentives may deal with downloading, but it won't have any effect on uploading?

6116   MR. DANIELS: I'm not sure -- and I don't know, Peter may want to elaborate on this point.

6117   The difference between us and Rogers is that their problem is solely on the upstream and ours is not. With the DSL network, ours is a downstream problem, primarily, but not exclusively.

6118   I will turn to Peter on that aspect.

6119   MR. VANDENENGEL: It will be hard to address Rogers specifically, but our counter, for example, counts both up and down in your monthly aggregate bill, so you could fashion a counter that would put less capacity on the upstream to help that mechanism work. It would be playing with how much you allocate for up and down.

6120   There could be a mechanism, but economics alone...

6121   I think that Rogers expressed yesterday that in order for economics to actually really have an impact, you would have to get the cap so low as to be punitive to the customer -- to the vast majority of customers.

6122   THE CHAIRPERSON: The reason I am posing the question is because Videotron, yesterday, suggested -- and they undertook to furnish numbers -- that, to the extent that they have economic discipline or incentives, and they have an effect on the downloading, they have a proportionate effect on the uploading, and that the two are basically, more or less -- they don't have to be, but it is their experience that if you throttle downloading, you also have the same effect on uploading.

6123   At least that's how I understood it, that their evidence so far suggests that, to the extent there is a change of behaviour by virtue of differential pricing, it is displayed both on the uploading as well as -- downloading, as well as uploading.

6124   MR. VANDENENGEL: I guess the notion is the behaviour of peer-to-peer, because there was that governance model.

6125   And I guess, if you are using more, you are going to be using more in both directions anyway, because the web is an interactive approach.

6126   If you are going to be using a lot of download in some fashion, you are communicating with other people, so you are going to be using a lot of upload as well.

6127   So managing the upload and download would be a matter of how much caps you set for each of those directions.

6128   THE CHAIRPERSON: No, I am posing the question because it surprised me. I thought that the two would be totally unrelated. The amount of people who visit your computer to upload -- I don't know why that would be proportional to the amount that you download.

6129   MR. VANDENENGEL: It goes back to, if you are going to use the internet, you are using it to communicate with family, so you are uploading photos, you are downloading photos, you are uploading your photos onto Facebook, and things like that.

6130   So there is a bit of interaction there.

6131   THE CHAIRPERSON: The key point that comes over and over again, especially with you -- and I don't want to revisit the CAIP decision, so let's stay on the generic rather than the specific -- is that wholesale ISPs who come to us -- your customers -- basically suggest that (a) they can control their network, and (b) to the extent that there is congestion, it doesn't pass up into your network and cause congestion. You have both the technological means to keep it contained and --

6132   A perfect example is Execulink, who says: If I have a problem, I deal with it. I can't pass the problem onto Bell.

6133   I asked them specifically that, and they said no, that they have technological ways to stop it.

6134   In effect, why, if you have network problems, do you have to pass them down to your wholesale customer?

6135   You address it in paragraphs 39 and 40 of your reply, and then today, again, you mentioned it in paragraph 15.

6136   Maybe you could explain it to me. You say here that it is commingled in the aggregation portion of the network.

6137   I'm sorry; I am not an engineer, so if could you explain it to me so that I, as a lawyer, could understand what you are talking about, I would appreciate it.

6138   MR. DANIELS: I would be happy to.

6139   Anticipating this question, in our opening statement we attached a couple of diagrams. Sometimes a picture speaks a thousand words. I am a firm believer in that.

6140   The first diagram that we have there is "Wholesale Aggregated ADSL- GAS Service".


6142   MR. DANIELS: Perhaps, lawyer to lawyer -- I will see if my engineer pokes me in a moment, but I will try to do this in a way that I can understand it.


6144   MR. DANIELS: What we show there is, starting from the right, you see a bunch of houses, and they are connected to what we call the CO, that is, the Bell Central Office.

6145   For simplicity, that is where the unbundled loop goes through. So if you want, like Primus does, to co-locate, you can come to our Central Office and you will get the unbundled loop from the Central Office. If you do that, you are dedicated to that individual customer, and you are never going to be on a shared network.

6146   Now, there are some exceptions to that, but generally that is the rule.

6147   From the Central Office, what happens is, we have to bring -- if you are buying our retail internet traffic service, or you are buying our GAS service, you are going to go through the Aggregation Network, and we are making that distinction -- this is how Bell has referred to it -- at some point in the network you come back to where we have what we call the BAS.

6148   Without getting into all of the details of what the BAS does, you are accumulating in a bunch of areas, and bringing it back to the BAS, a whole bunch of different COs, and it is there that we also have located our DPI. It is right there, behind the BAS.

6149   THE CHAIRPERSON: BAS being the Broadband Access Server?

6150   MR. DANIELS: Yes. That is where we do authentication, and so on and so forth.

6151   So it's at that point that you can view -- from there we have to go to the Backbone Network, and it comes back to the ISP.

6152   I am going to talk a little bit more from the ISP's perspective right now.

6153   If an ISP wants to buy the GAS service, what they get to do is, they get under our tariffs to interconnect at one point in Ontario and Quebec. I am talking for Bell and Bell Aliant, our Ontario and Quebec region. They can either interconnect at one point, or they can, if they want, interconnect at one point in each province.

6154   Let's, for simplicity, say that they interconnect at one point. They come and they tell us where, "This is the point where I want to interconnect with you," and they have to buy a different service from us, or whatever, but they come to that one point in our network, which is a Bell CO, located close to wherever they are.

6155   At that point they get access to every single end user that we can provide retail internet service to throughout our territory.

6156   Now, the nature of the GAS service is that it is set up to duplicate what we do with our Bell internet -- actually, it doesn't really duplicate, because it does it together. We pull all of that traffic, we take it at the Aggregation Network, and that's where we identify that it's wholesale traffic, and if it's wholesale traffic, we send it on the Backbone Network directly to that one location where the ISP interconnects with us.

6157   So, as you can see, there is a difference between the Aggregation Network, where that traffic is completely commingled, not separated, either virtually or physically, from our retail traffic, and then, when it hits the BAS, it is identified, that is wholesale, send it over -- that is the wholesale for wholesale carrier X, send it over to whatever location they have chosen.

6158   Now, one of the other things, while I have you on this diagram, to sort of complete the picture, is that the ISPs, when they interconnect with us -- we have to understand how the GAS tariff works.

6159   The GAS tariff has two pricing components to it. The first and most important pricing component is that they pay for each end user that they sign up, and the fees are as low as $19.50 up to about $22.50 per end user that they sign up.

6160   They also pay for something that you have heard referred as the AHSSPI, which is the total bandwidth. That is, at that CO where they interconnect to us, they pay a fee, on a monthly basis, for the total amount of bandwidth that they need.

6161   To illustrate this properly, what they are doing is just paying to connect into our port at that point. They are paying a fee for the port. That fee represents less than 2 percent of our total fee for GAS.

6162   If we looked at our top six ISPs who purchase GAS, and looked at how much they spend on a per end user basis at the $20 per end user, roughly, charge, and compared that with the amount of their AHSSPI payments, that other fee --

6163   The AHSSPI payment, just to give you an illustration, is $1,850 per month for one gig.

6164   That fee is only 2 percent -- less than 2 percent of the total amount of money they pay. That is because what they are paying for there is not access of a certain amount of bandwidth throughout our network, they are only paying for the door.

6165   We keep hearing references to ISPs saying, "We paid for the bandwidth," and that's not correct.

6166   So now that I have explained GAS and gone off on this tangent, I want to come back and answer your question. Your question to me was, "How is it that it's not segregated," and, as you can see, in the Aggregation Network, that's where it is not segregated and that is where we have our congestion problems.

6167   And then you said, "Is there a way that, if they treat the traffic, they can solve this problem by just doing it?"

6168   I guess that we have a couple of things to say in that regard.

6169   First of all, they do have an option to buy a service from us that skips the DPI. It is set up in the same manner. It is called HSA, High-Speed Access. It is priced higher. It is about double, it's about $40, roughly, at its cheapest. HSA, however, is designed -- it has some features that -- for one thing, it skips traffic shaping.

6170   So they have that option available to them, yet it goes through the Aggregation Network and Backbone.

6171   So they do have that option.

6172   But the second question really comes down to: Can they, by instituting whatever traffic shaping manner they would do, address their issue?

6173   Well, the first and foremost problem is we cannot -- although we can, it's all clear, turn it on for wholesale and not -- like we can turn it on for retail and turn it off for wholesale. You know that. We have all -- because actually that is what we did. But what we can't do, first and foremost, is turn it on only for one particular wholesale customer. It is either all wholesale or no wholesale.

6174   And in terms of could a different ISP institute a different solution, would it solve our problem, no, and that is why I am going to turn it over to --

6175   THE CHAIRPERSON: Is it a question --

--- Laughter

6176   MR. DANIELS: See, I did it. I stole -- we spent enough time talking about this that I feel that I have learned this.

6177   So Carl, anything -- nothing to add?

6178   MR. CONDON: I don't think so, Commissioners.

6179   THE CHAIRPERSON: Is it a question of the tariff that you have for GAS that you can't impose a condition on them, that they control their own traffic rather than that you do it on their behalf?

6180   MR. CONDON: Well, Commissioner, I think, to Jonathan's point, there is no way technically in the protocols to recognize the difference between the various wholesale customers.

6181   MR. DANIELS: We could, as a question of tariff, charge them more for the bases(ph) that they are going to take.


6183   MR. DANIELS: But we do do that and that is HSA. We give them the option.

6184   THE CHAIRPERSON: So if I understand it, looking at this diagram, these little houses here, et cetera, let's say just for argument, the top one could be one of your wholesale customers, an ISP that sells, someone like Execulink, and the others would be your own, they are all mingled up in the what you call aggregation network and therein lies the problem.

6185   MR. DANIELS: Correct, Commissioner.

6186   THE CHAIRPERSON: Okay. Thank you. I appreciate a non-technical explanation.

--- Pause

6187   THE CHAIRPERSON: Now, we have been asked by a lot of people to, in effect, not allow you to do the shaping or the throttling of traffic.

6188   What would happen if we actually did that? If we actually said, no, this, in our view, is a violation of 27(2), you can't do it, what would happen?

6189   MR. DANIELS: I am going to turn that over to my colleague Peter.

6190   MR. VANDENENGEL: I think it is clear we showed that peer-to-peer is a problem for us during peak periods. So if we were to turn off throttling, there would be increased instances of congestion. So removal of that shaping would lead to congestion, which would most likely lead to poor customer service.

6191   So from that point, in face of poor customer service, I would have to -- I can do two things. I can do nothing and leave my customers experience the congestion throughout the network, which I am not going to do, or I can invest, and if I invest, again, I can invest and have to redirect capital from other locations, from other network projects, or I would pass on potentially pricing increases to the customers to help me make up the money for that investment.

6192   So to follow my thought, if we were to eliminate shaping, which is one of our prongs, we would have to rely more heavily on the other two prongs, which is capital investment, and given capital is not unlimited, I would have to redirect capital and/or raise prices.

6193   THE CHAIRPERSON: But if it was across the board, what you are suggesting basically, the consumer would suffer?

6194   MR. VANDENENGEL: In one way or another probably, yes.

6195   THE CHAIRPERSON: Now, yesterday we heard from l'Union des consommateurs. There was a professor Mark Coates and he suggested in paragraph 28 that:

"In order to motivate the network management method of throttling a specific application, there should be clear measurement-based evidence that (i) there is congestion in the network; (ii) time-sensitive applications are experiencing unsatisfactory Quality-of-Service; and (iii) the throttled application is a significant cause of the congestion. There is little (publicly available) empirical evidence that these conditions exist. Utilization data cannot be accurately mapped to average loss rate or delay metrics, nor does it provide any information about the impact on time-sensitive applications. It is not clear from the presented data that peer-to-peer applications are the primary cause of the congestion, if it does exist."

6196   Now, he is obviously talking about public data. You have the data on your network. You can give it in confidence to us. Can you answer his question? Do you have the metrics? Can you show me that time-sensitive applications are or will be experiencing quality of service, that the throttled application causes significant congestion?

6197   MR. DANIELS: I would like to begin. The answer is yes and let's take you through some of the things that we can tell you and some of the stuff which is known.

6198   I think my colleague here Mr. Condon referred to the fact that P2P represents roughly a little over 27 percent of the total traffic on our network. So as a starting point, when people say it causes congestion, it is part of the problem that causes congestion but it is not exclusively responsible for congestion.

6199   The reason why we focus on P2P and traffic-shaping P2P is partly that it causes congestion but also because of how P2P operates in a congested state.

6200   If you would like me to explain that, I can. I can take you -- I have an explanation that can take you through that to explain what is unique about P2P in a congested state as opposed to just generally any other application in a congested state because it is there that it takes an unfair, disproportionate share of its bandwidth, in a congested state.

6201   So when people say, prove that it causes congestion, well, it is 27 percent or over 27 percent of our traffic, it is not the only cause of congestion but it is what happens in a congested state that becomes problematic with P2P.

6202   As for the question as to do we have the data of what the impact, I mean obviously we introduced it because of the impact on our users and so on.

6203   I am just going to turn to my colleagues to see --

6204   THE CHAIRPERSON: What I mean is can you take paragraph 28 and justify your present activities along the criteria that are stated.

6205   MR. DANIELS: Sorry, paragraph 28 of...?

6206   THE CHAIRPERSON: Twenty-eight of l'Union des consommateurs.

6207   I appreciate you don't want to share that data necessarily with the public but it would be very useful for me to be able to say, look, Professor Coates, we actually applied your test, we looked at it here and in our view we are satisfied that it applies in the Bell situation.

6208   MR. DANIELS: So would you like to take us through it now? Because we are prepared to.

6209   THE CHAIRPERSON: Fine. If you can do it on the public record, so much the better. Go.

6210   MR. DANIELS: Well, I mean, as a result of what we filed in the CAIP proceeding, I think a fair amount of this stuff is on the public record already anyway.

6211   There is congestion in the network. Well -- can you just give me a minute and we will get organized to give you --

6212   THE CHAIRPERSON: Sure. As I say, Mr. Daniels, if you prefer to do it in written reply in the next week, that is also fine with me. I just think, you know, this is a very legitimate question here and I would like to have an answer to it.

6213   MR. DANIELS: We are in your hands as to what you prefer.

6214   THE CHAIRPERSON: You can do it now by all means. Take your time.

--- Pause

6215   MR. DANIELS: While we are doing this, do you have our -- I know it was made part of the record of this proceeding. So do you have in front of you our final submission in the CAIP proceeding? Because that is where a lot of this data is found. I believe it is all on the public -- well, the key things are on the public record.


--- Pause

6217   MR. DANIELS: We are ready anytime you guys are.

6218   THE CHAIRPERSON: Okay, go ahead.

6219   MR. DANIELS: Okay. I am going to turn to the first part of the three-part answer, there is congestion in our network. We are looking at page 32 of our -- let's just be very specific for those who are following along -- July 11th, 2008 submission, our final argument in the CAIP, Figure 12.

6220   THE CHAIRPERSON: You filed that with this proceeding?

6221   MR. DANIELS: You made it a record of this proceeding, if I recall correctly, at the beginning.

6222   THE CHAIRPERSON: No, I don't have that in front of me.

6223   MR. DANIELS: Oh! You don't have that in front of you. We have a copy that we can give you if you prefer or we can do this written. It is whatever you prefer.

6224   THE CHAIRPERSON: Why don't you do it in written? I mean, as I say, Professor Coates asked this question. I would be interested in you being able to answer or justify your activities based on what he says here.

6225   And he also, on that same -- on the bottom he also suggests that deep packet inspection is not necessary, that you could do what you are doing on a shallow packet inspection, as he calls it.

6226   MR. DANIELS: If I could turn -- we will take that undertaking and answer that in this manner --


6228   MR. DANIELS:  -- but there are a couple of things that it would be better to explain orally.

6229   THE CHAIRPERSON: Please go ahead.

6230   MR. DANIELS: Okay. First off, in terms of the reference to shallow packet inspection, I am going to let my colleague speak on that in a moment, but as a lawyer, non-engineer, listening yesterday, this is what struck me about what was being said, is I understood the position yesterday was you should not allow application-specific traffic-shaping and then the further submission being made by the Union was you should -- shallow packet inspection would be sufficient.

6231   Now, what caught me is shallow packet inspection is just saying, don't use DPI, look at the traffic flows to see and recognize whether the traffic is a P2P traffic, and he rightly pointed out that part of the solution that even DPI uses is to look at the traffic flows to make that determination if it is P2P.

6232   If you are doing shallow packet inspection to make distinctions, you are looking for treatments, you are trying to recognize without going into the packet what application is involved. You only do shallow packet inspection to make distinctions on the basis of an application.

6233   So to say, don't do application-specific at the same time as saying, do shallow, there is no point in doing shallow. Shallow is just saying -- what the proposition was, Oh! You don't have to use DPI to recognize P2P. But if you don't want us recognizing -- if you don't want us making distinctions on the basis of P2P, why would you use shallow inspection? It doesn't tell you anything.

6234   THE CHAIRPERSON: I thought his motivation was privacy.

6235   MR. DANIELS: I agree to the extent -- all I was trying to say is that when we listened to that explanation it wasn't made in the context of saying, if you are going to do traffic-shaping, then you should do shallow.

6236   Now, as to whether shallow is sufficient or not, can we rely on it strictly assuming that you are going to distinguish by application, I am going to turn it over to my colleague Mr. Condon.

6237   MR. CONDON: The expert witness's concluding remark yesterday was that there is no commercial availability of that product and I agree with that. It is not a product that is on the market.

6238   THE CHAIRPERSON: Okay, thank you.

6239   MR. DANIELS: Mr. Chairman, there is one other thing I would like --


6241   MR. DANIELS:  -- not on the shallow but coming back to your question before. I said there is something that it is better to explain orally than just in the written form and it really goes to the question of how is it that P2P -- one of the questions that you pointed us to that we have taken an undertaking is to respond to how is it that P2P will affect real-time-sensitive nature of an application.

6242   For that, I really want to talk about the customer experience but specifically I have another graph, with your permission.


6244   MR. DANIELS: It is the last graph that we attached. So you see, you put these graphs in, you want to make sure you get a chance to speak to them.

6245   What this graph is meant to illustrate here is the difference between -- and this is based on something that Vaxination filed before and it is the argument -- it goes to the argument of saying, I have a 5-meg service, I am paying for a 5-meg service, it doesn't matter if I am using P2P or anything else, the most I can get is 5 megs. Intuitively that makes sense but let's bring some reality to it.

6246   The problem with the graphs that you have been presented is they have never shown what happens in a state of congestion. One would think how it should work in a state of congestion is -- so we have a before and after on the top.

6247   I am looking at scenario 1. Before, user 1 has 5 megabits, user 2 is using their full 5 megabits, and the total pipe available is 10 megabits.

6248   Now, imagine that all of a sudden there is a third user who comes in wanting to use their full 5 megabits. So now, we have got a situation where we have got 15 megabits of demand at that time.

6249   What happens -- what should happen notionally is each should get a third of the available bandwidth, in this case, you know, 3.3. This is just a notational concept to represent -- that is what you would think that each would happen fairly.

6250   But because of the nature of P2P, which opens a whole bunch of different sessions at the same point, which you can see in the bottom there, here are two users. One of them has got 10 steams open but is using their full 5 megabits and the other is using their full megabits for one stream, for example, doing video-streaming or downloading -- sorry, video-streaming at the time, YouTube.

6251   When the third customer comes in, user comes in and wants to use 5 megabits, its 5 megabits now, we are back on that congestion. What actually happens, because you have those whole sessions in, TCP/IP ends up -- just the nature of how congestion works -- giving, I mean not always exactly this amount but this is just a notational representation, the full 5 megabits stays to the P2P user but the other two users go down even more.

6252   That is why I say our problem isn't strictly with P2P as being -- because I told you it is 27 percent of our traffic -- it is not the sole cause but it is what happens in a congested state, it takes its unfair, disproportionate usage compared to other users.

6253   So you ask, how does that affect other time-sensitive applications? What it affects is user experience in terms of things that they want to do, video-streaming, time-sensitive, VoIP.

6254   So when you look at it from that perspective, what we are saying is it doesn't necessarily cause congestion, it is part of it, but because it is non-time-sensitive, we limit it to file-sharing and because it is the nature of time-shifting, what happens is that we take it down during our peak periods.

6255   Come 2:00 in the morning when we turn off our traffic-shaping, from 2:00 to 3:00 we see a big spike in P2P as the downloads complete because it is all machine-to-machine. So they are sitting there taking -- they are downloading but they have gotten part of their movie or whatever they are downloading. But come between 2:00 and 3:00 in the morning we see a big spike in P2P and it finishes downloading those movies available because people don't expect to watch them right at the same time.

6256   We find that this is -- it is the inherent nature of the protocol and by limiting it, and this is the key, limiting it only to file-sharing. We don't touch Skype, which uses P2P. We don't touch Juice, which uses P2P but is used on a more time-sensitive nature.

6257   By doing it in that manner, we can improve the overall user experience and the people who are using this are relatively unaffected. I mean it is not perfect. I am not suggesting it is perfect but they are relatively unaffected in terms of being able to get what they want and that is what is different about --

6258   THE CHAIRPERSON: Is this an explanation why you do downloading as well as uploading? All the others only do uploading.

6259   MR. DANIELS: The difference is --

6260   THE CHAIRPERSON: I mean the other people who testified here say they throttle the uploading. You, I believe, are the only one who is also throttling downloading.

6261   MR. DANIELS: In Canada --

6262   THE CHAIRPERSON: We are in Canada.

6263   MR. DANIELS: I make that distinction because there are many other DSL service providers around the world who traffic shape the download, BT in Japan. We have a whole list of carriers who can do it.

6264   You have heard from people who are cable companies. They have a different architecture and their problem is more geared towards the upload on the upstream because of the narrow bandwidth dedicated to that in their shared network. That is their issue.

6265   We have a different issue and I have a feeling that in a couple of years when wireless becomes -- when wireless broadband takes off like we all expect it to, I have a feeling that you are not going to see this upload/download distinction in the same way that you do today.

6266   THE CHAIRPERSON: Okay. So it is technologically driven? The reason that you also throttle download is because of the technology of our telephone network as opposed to cable network?

6267   MR. DANIELS: Absolutely.

6268   THE CHAIRPERSON: I am sorry, I have changed my mind, I am going to ask you to -- we will take a five-minute break. The document that you wanted to -- give it to the secretary so she can make copies for all the panel and then walk me through that. Okay?

6269   MR. DANIELS: We would be happy to.

6270   THE CHAIRPERSON: Let's do that. Let's take a five-minute break now while the secretary makes the copies.

--- Upon recessing at 0932

--- Upon resuming at 0942

6271   LE PRÉSIDENT : O.K., commençons.

6272   Let's resume, Mr. Daniels. I now have page 32 in front of me, which I gather is what you wanted me to look at.

6273   MR. DANIELS: And this time I am going to let my technical expert answer this.

6274   Go ahead.

6275   MR. CONDON: Commissioners, the first question on page 28 -- section 28, rather, of the submission was, does congestion exist?

6276   And in this diagram you can see escalating congestion. What's cell loss or packet loss, they're really the same thing, is when queues in switches or routers, in this case we're talking about ATM or a synchronistic transfer mode switches, but they're one of two technologies rather we use in the access network.

6277   When queues become overloaded, packets or cells, same thing, are dropped and require retransmission either by, in the case of TCP they require retransmission by the application which uses TCP to signal that that drop has occurred, or in the case of UDP, they're lost altogether.

6278   So, that was the first question.

6279   THE CHAIRPERSON: Excuse me, so it's lost and the system realizes it's lost and it won't proceed until it has resent it; is that how it works?

6280   MR. CONDON: If you're using TCP, which is transition control protocol, there is an acknowledgement, algorithm, Mr. Chairman, and it will ask the application to retransmit.

6281   The best example and most common example, a lost packet is not transmitted is a VoIP call and the application's mechanism, I suppose, for asking for retransmission is, I'm sorry, you broke up, can you repeat that, that's the application's mechanism.

6282   So, those packets in a VoIP call are lost forever in that sense, so they just get repeated.

6283   THE CHAIRPERSON: And that's what causes this pause in a VoIP call because the packet is lost and the system is waiting for it to be retransmitted?

6284   MR. CONDON: If it's lost --


6286   MR. CONDON:  -- and you say dropped part of a word, Mr. Chairman, you've lost it.

6287   It counts on -- if you and I are speaking on a call, but it counts on, sorry, not unlike a bad cell phone call, different reasons entirely, but the experience is similar.

6288   So, I would ask you, sorry, Mr. Commissioner, I lost -- what did you say after such-and-such.


6290   MR. CONDON: Could you repeat that, please. That is an example of packet loss somewhere in the network.


6292   MR. CONDON: I think I just explained No. 2, and then No. 3, so, is it the sole --

--- Pause

6293   MR. CONDON: So, is peer-to-peer a significant cause of congestion? It is a major cause of congestion. Every -- you know, if we just assume some equality in the packets for a second, every packet that's going through a device -- a congested device is a contributor to that congestion.

6294   In our case, as we said, 27 percent of our traffic is peer-to-peer, so it's certainly a significant contributor to that.

6295   And then I guess the question is, why do we handle it that way? And we go back to the diagram.

6296   Most of the traffic in the network assumes some immediacy of customer experience, and by that I mean that you're counting on getting that data pretty quickly. If I'm looking at the Tour de France results it's usually, though I don't expect them to come in say from the BBC in the next half a second, I expect them pretty quickly. I'm not going away.

6297   VoIP calls is another example of something like that, YouTube, things of that nature; whereas peer-to-peer is by its nature machine-to-machine and it's started and finishes at some point, and depending on the size of the files transferred it could be quite a long time.

6298   So, the issue is, how do you handle a customer experience at peak, and what we're in effect doing is prioritizing traffic that has a more immediate customer experience over machine-to-machine traffic.

6299   And that's in summary our approach.

6300   THE CHAIRPERSON: And the next --

6301   MR. CONDON: Oh, sorry.

6302   THE CHAIRPERSON:  -- Mr. Condon, the next sentence which says:

"Utilization data cannot be accurately mapped to average loss rate of delayed metrics, nor does it provide any information about the impact on time-sensitive applications."(As read)

6303   MR. CONDON: I would disagree with that. I've talked to many major telephone companies and we all use utilization data, we use that. I mean, the engineering people go looking at -- they look at queues in the devices and then -- but they all use utilization in the end because essentially it's a reasonable proxy to get large amounts of queue data -- real-time queue data from -- I told you I had 1,500 links. I made a mis -- for the record, 15,000 links in that network.

6304   To get that much real-time queue data isn't practical, but utilization is a reasonable proxy for that and it's used, as I said, by major telephone companies across the world.

6305   MR. DANIELS: And just the one other, just to take it back from a user perspective again, I mean, this is all saying they're not necessarily -- are they the cause of the problem? What we said it's part -- P2P is partly the cause.

6306   We don't have a problem with P2P, we're very happy for people to use P2P in our non-peak periods. And just to give you a sense of what the impact on our user overall, our top 10 percent of our users are responsible for 96 percent of our P2P, so traffic. Now, that's only retail. I don't have that breakdown and we can't do that breakdown in wholesale.

6307   So, looking at our retail traffic, the top 10 percent of our users are responsible for 96 percent of the P2P traffic.

6308   The rest, about half our users, use up about four percent. So, they're not using it very significantly and overall we believe that this is a user friendly way to do it because the vast majority of our users are getting -- we're doing it to improve their overall experience for all the other functions.

6309   So -- and I think Carl has described to you some of the problems of the impact on voice, or the impact on video streaming and so on, I think you have a good understanding of that, but to say that we can't actually prove that it has that impact, I mean, we're doing it because we believe in improving the user experience.

6310   And we're just time shifting when they download because the P2P time sharing is not time sensitive. So, again, I'm not sure I can add any more to that.

6311   THE CHAIRPERSON: Switching gears altogether, when you talk about privacy you're saying in paragraph 15 of today's presentation that the data is used for network management purposes. It is aggregated data.

"It is not being used for marketing purposes."

6312   And then in paragraph 23 you say:

"For example, data used for the purpose of ITMPs should not be used for any other purpose without customer consent."

6313   Surely this is a bit weak; isn't it? You're not allowed by law to do it, it's not only it is not being used, you're not allowed to use it for other purposes.

6314   MR. DANIELS: I'm going to hand that over to our privacy expert, Ms Morin.

6315   MS MORIN: Thank you, Mr. Commissioner.

6316   As we explained, what we're using DPI technology for is part of traffic shaping is only for those purposes.

6317   Privacy legislation allows us, if we wanted to use it for another purpose -- and that's an if, that's totally speculative -- we would have to go back to customers and seek their consent to do so.

6318   In the guidelines and in an effort to show reasonableness for ITMPs and disclosure to customers, we would do so under privacy legislation anyways, but we thought it would be reasonable to bring it because people seem to be concerned that maybe there were privacy implications, that sometimes just letting customers know that we're not using your information for any other purpose is a good thing to tell customers.

6319   THE CHAIRPERSON: So, you are telling me PIPEDA would allow you to use ITMP data, you know, DPI data for marketing purposes if you have the customer's consent?

6320   MS MORIN: We've identified a purpose for the data and it's traffic management only. So, internally that's all we've identified as a business purpose and marketing is absolutely not one of the purposes we want to use it.

6321   Privacy legislation --

6322   THE CHAIRPERSON: That's not my question. My question was very specific.

6323   MS MORIN: Yes.

6324   THE CHAIRPERSON: I thought you can collect this data for the purpose that you state which is ITMP, I didn't think you could use it for any other purpose.

6325   You're telling me you could use it if you had the customer consent? Not that you want to, I understand that, but you could.

6326   MS MORIN: PIPEDA is exactly based on that, it speaks about collection, use and disclosure with customer consent.

6327   I could give you an example.

6328   THE CHAIRPERSON: Okay. If that's the case, I'll take your word for it, you're the expert on that, but then you say you don't want to use it for that, you only want to use it for ITMP purposes.

6329   MS MORIN: Absolutely.

6330   THE CHAIRPERSON: Right. So, presumably then you would also have no objection if we made that as a condition, and we would impose that on ISPs, as saying any data that you collect for traffic management purposes can be only used for that purpose and none other, regardless of whether you have customer consent or not.

6331   MS MORIN: If the CRTC chose to make that determination, we would follow it.

6332   THE CHAIRPERSON: No, I'm speaking -- I'm talking to you as an ISP. As far as I can -- from your answer, I gather it would not cause you any problems?

6333   MS MORIN: At the present, however, you know, privacy legislation still exists as an overlay.

6334   THE CHAIRPERSON: Okay. And then, lastly, Mr. Daniels, your guidelines. And thank you for providing a set. I very much appreciate it because, as you gather, rather than making clear-cut rules, you know, that would be striking the balance between a free Internet and letting applications roam and be as innovative as possible and recognizing that you have needs for management and to protect your integrity, this is probably a good approach to go.

6335   But I am trying to understand why you are so insistent on a reasonable test rather than a least intrusive test.

6336   I mean, you started out from you want innovation, everybody wants innovation, you just want to have the ability to manage your network so that all users have maximum benefit, if I understand it.

6337   So, therefore, if there is any traffic management, shouldn't it by definition be least intrusive?

6338   Now, obviously there's an economic overload, there may be -- a least intrusive solution that is economically unfeasible doesn't make sense, but short of that I would have thought that least intrusive actually is what you want, it is consistent with the philosophy that you expounded in your submission and in your presentation this morning.

6339   MR. DANIELS: I guess the first thing -- I think what we're talking about really is, you asked me earlier, for example. different cable companies are doing different things in their network, comment on that, and I said I thought each one of them in and of themselves is reasonable, I didn't think anything was unreasonable.

6340   The problem I have with a test that says least intrusive is least intrusive inherently means that ultimately a judgment has to be made as to is there any better solution out there, because if there is, yours is not least intrusive.

6341   And take, for example -- in my mind the least intrusive implies more of a subjective analysis than the objective type of analysis that we were striving to do and that I've heard you looking for.

6342   And I say that because ultimately -- and I guess I'm repeating my words from my opening statement -- if the test is least intrusive to interpret as something -- unjust discrimination under section 27, what's going to happen is people are going to come in and say, oh, we think you can do it this way.

6343   We have someone who comes in and puts in and points to, oh, there's a vendor out there who's advertising a flow management, we think that's least intrusive, without even knowing whether it's commercially deployed or so on, without knowing the nature of the network.

6344   And ultimately what happens is with the least intrusive test, I think the Commission would have to decide looking at it and say, is this one less intrusive than the other one; if it is, boom, that's the one you have to do and I think it requires the actual -- your determination as to what is the best method.

6345   That's my problem.

6346   THE CHAIRPERSON: Okay. Let me stop you there because --

6347   MR. DANIELS: Yeah.

6348   THE CHAIRPERSON:  -- in Essential Services we made -- we talked about what's essential and we looked at it basically whether it's economically and technologically feasible, something like that surely would be here too.

6349   I mean, if you have to determine what's least intrusive or not, presumably would be on you to demonstrate here, this is what we can do, the other things are either too expensive or else they are technologically not feasible, and you would do that on the basis of data.

6350   So, where does the subjectivity come in? We would look at it at the same way as we did look in the Essential Services, et cetera, at the very service, but it's not a subjective decision, it is, you know, it is a reasonable decision based on evidence brought before us.

6351   MR. DANIELS: I think -- I mean, I appreciate that it would be before evidence brought before you, but ultimately in the Essential Facilities decision, for example, the Commission made the determination as to what was an essential service and what wasn't essential and every service, it was ultimately yes or no here it is, for all -- not all time, but for all players in the market or for the cable companies, here's the ones for you, for all ILECs, here's the ones for you, it was an actual determination.

6352   And I think the best way maybe to think about this is, to think about it from how the law and a judge looks at the question of reviewing an administrative tribunal decision.

6353   There's a decision between a correctness standard and a reasonableness standard. And when the regulatory agency is in its area of expertise the court looks and reviews in facing an application for judicial review, generally applies a reasonableness standard which says, look, I'm going to look determine based on the evidence you filed, did what you make that decision, was it reasonable based on the record of the proceeding.

6354   It's not necessarily the one I would have chosen, it's not necessarily the right decision in my mind, but was it reasonable because you -- and I'm referring to the court says -- the regulatory agency has better expertise in this area.

6355   Contrast that with correctness. This is an area where I as a judge, I know better or equally as well as the regulatory agency, it's interpreting a statute, for example, outside of its jurisdiction, on that point my view is the one that I as the court applies.

6356   And I say that's the same parallel. I'm worried that least intrusive in terms of -- in this context here that least intrusive would become, if applied by the Commission, would become sort of similar to a correctness standard which is to say, ultimately it's your view that matters as opposed to saying, I'm going to defer to those who have better expertise, but not complete deference, I'm going to look and determine, did you guys choose a method that is reasonable in the circumstances to limit the negative impacts on the users who are affected.

6357   I hope that tries to --

6358   THE CHAIRPERSON: No, I understand. You are obviously playing to my past, you know the source of the concept, I am very familiar with.

6359   But, I mean, this also will be ex post, not ex ante, right. You would come forward, you would -- if you do it on an ex post basis, I think the fear that you have that CRTC lays out, in effect, like in Essential Services yes or no, is probably to some extent alleviated if it's an ex post facto review.

6360   MR. DANIELS: It certainly is alleviated to some extent if it's ex post. Clearly I know there has been some discussion about having application-specific ex ante and we would strongly urge you that we are opposed to that.

6361   But perhaps I can give you sort of -- like let's just take a look in the cases that people have brought before you. There has been reference for example to protocol agnostic solutions. I'm not going to even -- I'm happy to get into about the problem of protocol agnostic solutions in the telecom space as opposed to the cable space, but people -- in terms of product being out there in the market.

6362   But even if such a product exists -- and it doesn't today that we are aware of -- even if that solution did exist, the problem with protocol agnostic that some people have pointed out is that it impacts if you are -- the way the Comcast new protocol agnostic solution works is it focuses on people who are using applications, at and any one particular time if they have gone up to the 70 percent threshold, for the next 15 minutes the slow down all their traffic, including their time-sensitive traffic.

6363   Some people come before you and say we think that's least intrusive because it has the benefit of not distinguishing between applications and treating all applications equally, and that's their view. Frankly, I can understand that. I'm not trying to say that a protocol agnostic solution is unreasonable. However, it has the impact of affecting someone who wants to do gaming or Skype or something at that time.

6364   We have come up with a solution that targets application specific P2P, what we -- non-time-sensitive traffic at a different period of time. Is it less intrusive than a protocol agnostic solution?

6365   Some people are going to come before you and tell you unequivocally -- I think this is the OIC -- unequivocally absolutely it is less intrusive. I'm not sure I agree with that. I think ours is probably less intrusive to the overall user experience.

6366   Having said that, it's ultimately the arbiter's decision. And that's why I'm saying, if the test is least intrusive, you have to choose and make that decision, but if the test is reasonable you don't have to make that decision. You can say "Bell, what you are doing, it's reasonable in the circumstances."

6367   I'm not --

6368   THE CHAIRPERSON: Okay. Well, would say it's not unreasonable; right?

6369   MR. DANIELS: Sorry?

6370   THE CHAIRPERSON: Is not unreasonable. That's really what your whole --

6371   MR. DANIELS: Yes, exactly.

6372   THE CHAIRPERSON: That would be the test.

6373   MR. DANIELS: Exactly. That's what I'm really trying -- exactly the distinction I'm trying to make.

6374   THE CHAIRPERSON: Okay. Well, thank you very much.

6375   I will pass on to my colleagues.

6376   Len, you have some questions?

6377   COMMISSIONER KATZ: Thank you, Mr. Chairman.

6378   Good morning. I'm going to go back to the discussions you had earlier on with the Chairman.

6379   Firstly I just want to get some facts clear in my own mind.

6380   First of all, your traffic, you have identified that your proposal here is directed at the public Internet and not at the managed network VoIP or anything else as well.

6381   Do these networks coexist in a breathing environment or are they so totally independent of each other that any traffic or congestion on one doesn't flow over and impact on the other?

6382   MR. DANIELS: I am going to let Carl answer that question.

6383   MR. CONDON: Yes, Mr. Commissioner, they are all in the same pipes and they are all best effort.

6384   COMMISSIONER KATZ: So that if your managed enterprise network, IP platform and everything else as well, has got a high incidence of traffic, it will congest the public Internet?

6385   MR. CONDON: Let me give you a broader answer, Commissioner.

6386   In our network there are multiple traffic types. Some of the business services do have a higher priority than some best effort services, whether they be business or consumer, and they are treated with priority through the network. If we didn't engineer the network correctly, yes, there would be congestion.

6387   I think you asked a question that perhaps you are going to come around to, IPTV as well, and the architecture for IPTV is to build out -- you would run an IPTV network in -- ITV, sorry, solution or service in a best efforts Internet. What you would end up doing -- and this is done around the world -- is you actually augment the capacity you have -- because you never have enough capacity in a pure Internet-only backbone for IPTV -- and then you then guarantee that traffic.

6388   So let me give you an example. If you are perhaps building a high occupancy or a carpool lane across the 401 through Toronto, you might build those lanes on, and that's the equivalent of adding capacity for IPTV. That is reserved for IPTV, but when IPTV is not using that capacity it is available to anything else.

6389   COMMISSIONER KATZ: But when IPTV needs more capacity it's going to take some of the lanes of the highway?

6390   MR. CONDON: I wouldn't say that, Commissioner. I think if you are going to offer an IPTV service you are going to have the right amount of capacity built and engineered for IPTV. You wouldn't underbuild it.

6391   COMMISSIONER KATZ: When TELUS was here on whatever day it was, Thursday of last week --

6392   MR. CONDON: Right.

6393   COMMISSIONER KATZ:  -- I asked the same question of Mr. Hennessy and he said where there is congestion he believes that TELUS would likely not want to impact the public Internet because they are his customers I guess, not that any other customers aren't his customers as well.

6394   Did you see that testimony?

6395   MR. CONDON: I was present for that, yes.

6396   COMMISSIONER KATZ: Okay. Was I correct in what I thought I heard him say?

6397   Put it this way, why don't you take a look at again and why don't you, in your final response, tell us whether you agree with his views or not.

6398   MR. CONDON: Yes, we will do that. But I think I would agree with your statement that in IPTV architectures you don't diminish the capacity available to the general Internet.

6399   MR. DANIELS: The way we have been viewing it is sort of like you build that additional lane and during -- you know, it's like in the lane in the highway, they add an extra lane for the carpool lane and then in hours when the carpool lane, when it's not -- you know, that's only used Monday to Friday so on the weekend all traffic can use all aspects of it.

6400   So we actually think what you have to do is you have to provision enough capacity, give the IPTV what it needs and then during the off-peak periods of IPTV when it's not being used, that actually --

6401   COMMISSIONER KATZ: I was asking the reverse.

6402   MR. DANIELS: No, I appreciate you are asking the reverse in terms of will the IPTV start to shrink the actual --

6403   COMMISSIONER KATZ: That's right.

6404   MR. DANIELS: I mean, I think if you can see what's going on in the competitive market, we are not going to be in a place where we are going to be interested in trying to decrease speeds in our experience on the Internet. It's part of our solution.


6406   You also said, I believe, that you are able to distinguish wholesale from retail as you have done as well, but I think I heard you say you can't distinguish the applications on wholesale necessarily.

6407   MR. DANIELS: No. What we do for wholesale is exactly what we do for retail, so the distinction between P2P is exactly the same, there is no -- what I meant to say, if I didn't get it across, is that when one party like Execulink comes here and says, look, I can manage my own network, I will take care of any problems that you are facing, so don't shape me, I said I don't -- I have 100 ISPs on my GAS service, I don't have the option of distinguishing between each of those hundred ISPs. I can either turn on shaping for all of the 100 ISPs on GAS or I can turn it on for none of the hundred ISPs. I am not able in my network to distinguish and apply it to one specific ISP, have an exception for one, not for another, have different rules.

6408   We have to, in the aggregation network, treat everyone exactly the same and, quite frankly, if someone came to us, like Primus for example, and said "Oh, don't do it, we have this different type of priority". You know, they talked about their application-specific -- and I do want to correct the record or make sure that it's understood that Primus is talking about separating three different time sensitive, non-time -- you know, more time sensitive, I can't remember the exact language and then best efforts. That is application-specific distinction because they are looking at applications and slotting them in.

6409   COMMISSIONER KATZ: Just on that issue before I forget it --

6410   MR. DANIELS: Yes...?

6411   COMMISSIONER KATZ: Primus does use a DSL technology.

6412   MR DANIELS: That's correct.

6413   COMMISSIONER KATZ: Primus claims that they have been able to traffic manage and do some of the things that you, I think, said earlier were not able to achieve.

6414   I'm just wondering whether your networks are the same or not.

6415   MR. DANIELS: I'm not sure we said that. We aren't -- I don't think that they said that there is something that we are not able to achieve.

6416   What they do -- just give me one second because I want to -- I feel that I should be very specific if I'm talking about another party and I have it here in terms of exactly what they do.

--- Pause

6417   MR. DANIELS: They say not when they are using our GAS service, but when they lease, through their partners or whatever, unbundled loops from us.

6418   They make a distinction between very time-sensitive traffic such as VoIP and gaming. It's classified as expedited, the highest priority level.

"Time sensitive traffic such as e-mail, instant messaging, streaming media is classified as high priority. Traffic that is less time-sensitive, including P2P traffic is classified as normal priority."(As read)

6419   COMMISSIONER KATZ: But they are doing that, I think they said, during congestion periods only. And so they are able to distinguish when there is congestion and when there is not.

6420   I thought, and I believe you said here in your submission, that you provide what you call narrowly focused traffic management, but yet the record shows that you start it at 4:00 p.m. and go until 2:00 a.m. in the morning, which is 10 hours which, quite frankly, I don't think it's narrowly focused at all.

6421   MR. DANIELS: The first aspect, I don't recall whether Primus -- I mean I could stand to be corrected, but when I'm reading their interrogatory response here, I don't recall them saying that it's only during --

6422   COMMISSIONER KATZ: They did say that. I will let you go back to the record yourself.

6423   MR. DANIELS: Yes. No, I will have to look at that aspect.

6424   COMMISSIONER KATZ: They didn't say that. It was only during periods of congestion they said and after that it's all alleviated and they remove all the shaping or whatever as well.

6425   MR. CONDON: Commissioner, we will go back and look at it. I don't recall it either.

6426   But I don't know what their definition of "periods of congestion" are, whether they are broad, whether there is -- I'm not aware of any signalling mechanisms in the DSL network that will allow them to understand real time that the congestion is occurring.

6427   They likely have a much, much smaller network than ours and perhaps they are manually watching that and turning things on and off manually, but I'm not aware of any signalling mechanisms that will allow them to do that in any automated fashion.

6428   COMMISSIONER KATZ: Okay. But in your paragraph 11 of your statement this morning, under bandwidth management you say:

"The practices we implement our narrowly focused..."(As read)

6429   And yet I believe, unless you can correct me, you are traffic managing from 4:00 p.m. until 2:00 a.m., seven days a week, 52 days a year.

6430   So how is that narrowly focused?

6431   MR. DANIELS: Well, I guess it's narrowly-focused in a couple of aspects.

6432   First of all, just to be specific, 4:30 to 6:00 we do a certain amount of traffic shaping. From 6:00 to -- one?

6433   MR. CONDON: One, yes. 6:00 to 1:00.

6434   MR. DANIELS: 6:00 to 1:00 we take it down. So from 4:30 to 6:00 P2P is restricted on an end-user two 512 kilobytes per second; from 6:00 to 1:00 it's down to 256; and then from 1:00 to 2:00 it goes back up to 512.

6435   So the first aspect is that we don't just -- we take it down, as we get closer to the high point of the peak we make a distinction, so it's just not totally on our.

6436   The second aspect, we think it's narrowly focused in the sense of -- I'm sorry, I can't read your writing.

6437   MR. CONDON: Transfer.

6438   MR. DANIELS: The key aspect is that it's limited only to non-time sensitive P2P traffic and so it's only filesharing. And there are lots of allegations out there about, you know, impact on non-time-sensitive or on time-sensitive applications and there is no evidence to prove it. There are allegations, but there is -- every time we look into this there is a host of reasons people may be facing problems, but there is no --

6439   When we investigated, every single circumstance, with the exception of one, it hasn't been as a result of -- we haven't had any collateral damage, to use the term, as the impact on non-time-sensitive applications.

6440   COMMISSIONER KATZ: The speeds you just quoted that you take it down to during peak periods, congestion periods --

6441   MR. DANIELS: Yes...?

6442   COMMISSIONER KATZ:  -- is that public knowledge? Is that readily available to any consumer or customer of yours?

6443   MR. DANIELS: We have put it in the submission and then we are going to be putting it on the website.

6444   So the details around it, but the actual speeds weren't on our website, but they will be on our website.


6446   MR. VANDENENGEL: Just to be clear, we have a lot of details on our website on our traffic management practices and the speeds was the only thing missing, so we are going to be adding that to the website.

6447   COMMISSIONER KATZ: Can I refer you to your paragraph 16 of your opening comments this morning. Maybe I was hypothesizing as opposed to listening, but where you said here towards the end of it that:

"...wholesale GAS traffic accounted for 31% of our total GAS and retail traffic. Further GAS traffic accounted for 36% of out total P2P traffic."

6448   Did you add after that another statement that's not here about your total traffic?

6449   Can you repeat that statement, because I thought I heard it.

6450   MR. DANIELS: Sure. Basically what I'm saying is that whereas GAS represents 31 percent of our traffic the GAS end-users represent roughly half of that. It's actually a little less than half of our total end-users of our total Internet end users.

6451   COMMISSIONER KATZ: So 31 percent of your customers use GAS traffic, of wholesale customers, but your total wholesale customers consume 50 percent of your total capacity?

6452   MR. DANIELS: No, no, no. So let me be more specific.


6454   MR. DANIELS: It's about 15 percent of end-users, of our total end-users, are GAS end-users.

6455   COMMISSIONER KATZ: Fifteen percent.

6456   MR. DANIELS: Fifteen percent, but they represent 31 percent of our total traffic.

6457   COMMISSIONER KATZ: Okay. Okay. So they consume a disproportionate share of your total traffic you are saying.

6458   MR. DANIELS: They do.


6460   MR. DANIELS: But let's be clear, it's generally not a problem. It's only a problem during periods of congestion and we have taken issues treating our wholesale and our retail in exactly the same manner to address that issue.

6461   All I'm trying to emphasize is we are trying to say that they have come forward and said we are so insignificant we can't be impacting the network.

6462   In our case that's not true. Perhaps in Rogers who came forward and said to you we don't traffic shape wholesale because they are just, you know implied that it's insignificant, but if they grow they will have to address it. We had to address it.


6464   You went over this wholesale aggregated ADSL GAS service charge and I heard you say that for those customers who want to they can buy an HSA service instead and that would bypass the best efforts basis that's available because they get a dedicated pipe effectively.

6465   MR. DANIELS: I didn't say that it would bypass the best efforts. What I said is that they would bypass the traffic shaping, the DPI.


6467   MR. DANIELS: This is where if you want a better understanding of this I'm going to turn it over to my colleagues.

6468   COMMISSIONER KATZ: No. I don't need technical solutions at this point in time.

6469   You said just now that 15 percent of your customers purchased wholesale GAS traffic.

6470   Are you at liberty of telling us how many pick up the HSA traffic service?

6471   MR. DANIELS: I don't have that number readily handy right now, but I know it's not terribly -- in terms of the total Internet --


6473   MR. DANIELS:  -- it's not a significant sum. It's a very, very small --

6474   COMMISSIONER KATZ: Could you file that with us as well?

6475   MR. DANIELS: Yes. We can take an undertaking, but it's small.

6476   COMMISSIONER KATZ: Okay. And I will follow up with a question. I will follow-up with a question anyways.

6477   One of the reasons why that tariff is more expensive I would imagine is because there is more cost associated with some of the components because you are dedicating Q of S to that.

6478   Is that right?

6479   MR. CONDON: No, Commissioner. It's because it is not -- it is a best efforts service through the network, it uses the same aggregation network and it bypasses the backbone network. It is a nailed up connection, a permanent virtual circuit, from the customer's premise, business or consumers, probably more often business service.

6480   COMMISSIONER KATZ: But there is no traffic shaping?

6481   MR. CONDON: There is no traffic shaping on it. It's nailed up from the -- nailed up, as I said, right through to the aggregated high-speed service provider interface, the --

6482   COMMISSIONER KATZ: And it's at a higher price?

6483   MR. CONDON: Yes, it is.

6484   COMMISSIONER KATZ: Let me hypothesize this to you. If in fact there wasn't a GAS tariff at all available to wholesalers but only an HSA tariff, there would be much more demand and much more volume for that service.

6485   To what extent would the price of that service come down because the volumes are so much more?

6486   Because obviously the HSA tariff is based on a certain forecast of demand and therefore it's priced to recover that and return some investment of shareholder value to you guys as well. But at the end of the day, if it's only one service and it's a guaranteed service as well, without traffic shaping, not guaranteed from the Q of S but from a traffic shaping perspective, there would be much more volume, presumably you can distribute your costs over a bigger base and therefore the rate would come down dramatically.

6487   I think the question is: How dramatically?

6488   MR. DANIELS: I'm not -- you know what, I'm going to talk to my colleagues on this.

6489   My gut reaction is I don't think it would come down, but anyway hold on.

--- Pause

6490   MR. DANIELS: I'm going to hand it over to my colleague, Mr. Condon, but I can confirm that I was correct in my assumption that actually it's not going to bring the price down, it's the opposite impact. He will give you an explanation of why in terms of what actually happens in the network from it.

6491   But just think about the fact that at that point in time we are going to have that much more traffic in our aggregation network that is going to be using P2P and grabbing the congestion points on it. So the end result of that is going to have to be that we are going to have to configure our network differently in order to accommodate that, to give that preference to wholesale. So that's going to cost money.

6492   But over to Carl.

6493   MR. CONDON: What that entails that doesn't exist in the GAS service, Commissioner, is it has to be provisioned through the entire aggregation network and bypasses the backbone network all the way up to the AHSPI.

6494   So all of the elements have to be individually provisioned, have to be inventoried, customer records set up and, as I said earlier to Jonathan's point, it would consume more capacity in the aggregation network than a current GAS service.

6495   So it's a considerably more costly service to build than GAS today.

6496   MR. DANIELS: The other aspect, I mean, ISPs generally by our GAS to compete with our retail products and the cable companies retail products. I don't think as a solution it's available if they want to have a premium to say "Oh, if you are a high P2P user come to us, we have a product out there but we are going to charge more in the market."

6497   But, you know, if we had such a ruling and such a tariff, and so on, and that GAS didn't exist, I know that there is going to be complaints about the difference between the retail price and what would be offered. How can we compete with you? Well, it's not the same service at that point in time.

6498   So isn't it much better to give the ISPs the option and have the rates that reflect that?

6499   COMMISSIONER KATZ: Do you offer end-users an HSA equivalent and service, retail customers?

6500   MR. DANIELS: In the residential market no. In the --

6501   MR. CONDON: It's widely used in the business market, Commissioner.

6502   MR. DANIELS: I mean it was created primarily for business, but we do have lower rates for high-speed access in the residential market, but we do not offer it ourselves as a retail. So this is an advantage that a wholesale provider can do to distinguish their service if they want from our retail products.

6503   COMMISSIONER KATZ: But if I'm a residential customer, I could buy the business service if I wanted to, I just have to pay the business rate I would imagine?

6504   MR. DANIELS: Yes. Yes.

6505   COMMISSIONER KATZ: Thank you.

6506   Those are my questions.

6507   THE CHAIRPERSON: Candice...?

6508   COMMISSIONER MOLNAR: Thank you.

6509   Since we were talking with Commissioner Katz about GAS I'm going to just finished with my question on GAS at the same time.

6510   I went back to the minutes of our conversation with Execulink earlier in the week and they make the point that there are the two components and you spoke about the two components, the access component of GAS as well as the bandwidth component of GAS. You commented that the bandwidth component is only, what, I think you said 2 percent of the entire price.

6511   MR. DANIELS: Less than 2 percent.

6512   COMMISSIONER MOLNAR: Less than 2 percent.

6513   But I would just like to understand because -- and again I'm going to use our conversation with Execulink as the example, where they said -- in their example they said that we will provide each customer up to 11 megs and that we use up to 100 gigs of traffic. Their bandwidth is 100 gigs.

6514   And their comment is that when you throttle the P2P that they slow it down at the customer. We don't get to 100 capacity because they have slowed down at the customer's end.

6515   Is it the case that by throttling the P2P through your peak period you have affected the total amount of capacity going over the wholesaler's network?

6516   MR. DANIELS: Let me answer that by saying that certainly by lowering the P2P, if their customers -- it really depends, every individualized fee is different. If their customers are not going to do other things with that extra bandwidth -- I mean the reason we take it down is so that end-users can do video streaming and so on. In terms of taking that down, does it impact? It would have to be the nature of their end users. If they had end users that were mostly P2P users, who don't do other applications, then, yes, our activities could impact the total amount of bandwidth that they are delivered, or take back in their network.

6517   If, on the other hand, it allows their users -- for example, our own retail P2P customers or their own -- if they have very small P2P users on their -- that are actually using it, they could see a greater amount of bandwidth.

6518   But the one thing I can tell you from a factual perspective is, I spent time looking from our wholesale -- I was talking to our wholesale team -- because last year a big deal was made about this issue. "We bought a certain amount of bandwidth," which they didn't, and I can explain that in a minute, but, "We bought a certain amount of bandwidth," said the ISPs, "and now you have taken us down, and we don't need that total amount of bandwidth."

6519   We have seen nobody -- we have seen no correlation between our traffic shaping and reduction requests in terms of AHSSPI connections that people have looked for us. We have seen the same typical growth.

6520   And we have also seen that year-over-year growth of GAS end users was the same from 2007 to 2008 -- February of 2007 to February of 2008 -- before we instituted traffic shaping, as it was the year after.

6521   COMMISSIONER MOLNAR: Can I interrupt?

6522   MR. DANIELS: Yes, absolutely.

6523   COMMISSIONER MOLNAR: You said that you have seen no difference between what -- after you instituted the traffic shaping, with the amount of bandwidth that they purchased?

6524   Is that what you said, or the amount of bandwidth that is being transported to the wholesaler?

6525   MR. DANIELS: I don't know about bandwidth that is being transported, because --

6526   COMMISSIONER MOLNAR: Do you know -- and that was part of my --

6527   MR. DANIELS: No, I can't --

6528   COMMISSIONER MOLNAR: You don't know?

6529   MR. DANIELS: All I know is whether they came -- going back to our diagram, whether --

6530   Look at it from an ISP's perspective. When they come to us and they need to interconnect, they are saying: I am going to have X amount of end users. I will pay you for every end user when I set up, but I need one pipe to connect to for your whole network.

6531   At that one point in time, we need to size the door. They choose the door size at that point in time.

6532   So all I can tell you is that we have not seen a decrease -- in fact, as they continue to grow their business, because they are having good growth, we have seen an increase in requests for the door size.

6533   Does that mean that there might be slightly less traffic coming through the door that they chose? Possibly. That really depends on the nature. Other ISPs, because of our shaping, might have seen greater.

6534   COMMISSIONER MOLNAR: Is there anything within the tariff that would limit them -- if they needed a smaller door size now because their traffic has been shaped in the peak period that they likely provisioned their door to work with, is there anything that would prohibit them from moving down to a lower size and getting a different capacity?

6535   MR. DANIELS: No, I don't believe so, and, as I say, factually we have not seen it.

6536   COMMISSIONER MOLNAR: You have not seen it.

6537   MR. DANIELS: In fact, we have seen requests for more doors, and they are signing up more end users, so we haven't seen that at all.

6538   MR. CONDON: I think we can say, Commissioner, that we saw continued, steady orders of AHSSPI post --

6539   COMMISSIONER MOLNAR: At similar sizes -- similar or greater?

6540   MR. CONDON: They increased, they didn't diminish.

6541   The capacities weren't diminished, they were increased.

6542   MR. DANIELS: But that is not to say that any one individual ISP may not have changed and gone down. I am talking trends, and we have not seen --

6543   COMMISSIONER MOLNAR: But, ultimately, if your traffic practices would change the amount of capacity they required, they would have the ability to change their order and change their costs accordingly?

6544   MR. DANIELS: Yes, I believe so, in terms of AHSSPI.

6545   Also, there is one other thing to note. The AHSSPI that they order they use for GAS and many other services.

6546   I mentioned that it is less than 2 percent, and that is assuming it is only used for GAS. But, in fact, it also has the advantage --

6547   All it is -- it's the door for all of the services that we take throughout the network, and the reason why I am focusing on the difference between a door as opposed to a pipe is that ISPs are trying to suggest to you that, because they buy a total bandwidth, it gives them a pipe of bandwidth throughout our whole network. It doesn't. It is priced strictly for the door, right where they interconnect, because it gives them access to our whole province on the shared network.

6548   COMMISSIONER MOLNAR: Would you know if that's the same way it is priced in other territories, for other ILECs?

6549   MR. DANIELS: I know that in the Atlantic provinces, Bell Aliant -- it is one place, one connection, that gives you all four provinces.

6550   We didn't look at the -- I don't think we looked at the AHSSPI.

6551   However, Manitoba, MTS, comes -- you know, our rate, for example -- 1 gig is between $1,750 and $1,850. In Manitoba, 1 gig, that one door, that setup is priced at $68,500.

6552   And they put in an application to you, to the CRTC, which has since been rolled into the ADSL proceeding, arguing that they provision it giving the bandwidth throughout the network.

6553   We don't understand even how they could do that for $68,500. We think that part of the answer, quite honestly, is that -- we can't wait to find out how many GAS customers they have, because we haven't found one.

6554   But they have obviously -- you know, what is that -- 3,000 percent rate on top of the -- as compared to ours.

6555   So maybe theirs is structured differently. That's not how our GAS is structured.

6556   COMMISSIONER MOLNAR: It sounds like it may be structured differently. The nail-up, as you were talking about it -- they spoke of a virtual private network versus the shared.

6557   But that's okay, we will have different periods to look at that.

6558   I wanted to ask you a little bit about your comments. In paragraph 31 of your comments today you spoke about guidelines that are flexible enough not to restrict network innovation, while protecting application innovation.

6559   Would you view something like DPI for purposes of traffic congestion to be a network innovation?

6560   MR. DANIELS: Absolutely.

6561   Let me give you an example of some of the types of network innovations that exist.

6562   I do think that DPI is a network innovation. It is allowing, at the network layer, to be able to improve the overall user experience.

6563   Blocking spam, for example, is a network innovation. It is done at the network layer. We go in and we pull out spam.

6564   And we think, for example, that we do that better than other companies. Now, that's a claim that's hard to establish, but I know that, as a Bell internet subscriber at home, I don't get spam. I know that other people do.

6565   So I consider that a network innovation.

6566   And if you look at what is going to happen on the wireless side, it is built right into the international standards, for example. In HSPA -- there are four classes of service associated with HSPA.

6567   When you go to LTE, which we are all talking about, in a few years' time, at that point voice is not distinguished in terms of -- it's treated as an IP in LTE, and at that point there are --

6568   LTE standard has nine different classes of service that are associated. For example: Priority 3, real-time gaming; Priority 5, non-commercial video; Priority 6, conversational video. Built right into the standard, and that, clearly, in my mind, is a network innovation.

6569   So, yes, I do believe that our practice is network innovation, and I think that network innovation is a very important point.

6570   COMMISSIONER MOLNAR: Early on in this hearing we heard from Sandvine and Juniper, and Juniper spoke about the potential that congestion management could occur with a customer; whitelisting of services, and so on, could occur at the customer, instead of being a choice of a network provider and you making the decision as to what is a priority.

6571   Do you have any comments on that?

6572   They did mention that it is nowhere in Canada today, but as a large network provider, have you looked at those types of options? Is it something you are looking at?

6573   You were speaking of LTE and so on. Is it the evolutionary plan that those sorts of controls can be held with the customer, or is that something that you would see designing into your long-term evolution as a competitive differentiator?

--- Pause

6574   MR. DANIELS: I'm sorry; we are just having a discussion as to which one of us will answer that question.

--- Pause

6575   COMMISSIONER MOLNAR: If you have to choose between the engineer and the marketer as to who should answer this question, I think you have half answered it.

--- Laughter

6576   MR. DANIELS: The nature of the question -- and I was turning to Carl to see if he wanted to specifically talk about something with respect to the Juniper solution, as opposed to talking to Peter about what we want to say about giving that to customers.

6577   The short answer is, let the lawyer in the middle --

6578   COMMISSIONER MOLNAR: No, we don't want lawyers doing this.

--- Laughter

6579   MR. DANIELS: Well, then, I will let Peter answer that aspect.

6580   There are a whole bunch of different, reasonable solutions that are out there, not necessarily all practical at this point in time.

6581   Total user control, I think that could be something that could come.

6582   Peter, I don't know if you want to elaborate on that.

6583   MR. VANDENENGEL: Sure.

6584   Just for the record, I am not a marketer, I am a product manager, and I really care about the customers more so.

6585   From a whitelisting point of view -- you asked if it was the engineer or the marketer who was going to answer the question, and it's also a question of -- whitelisting -- is Bell going to manage the network, or are we going to leave that in the control of other folks.

6586   From a whitelisting point of view, I think it has potential. We are always looking at ways to augment and make money off of our network.

6587   I don't think it is actually something that is going to be viable in today's market, and I don't think it is something that we could actually implement anytime soon.

6588   I think what we are doing now works for us, and that's all I have to say, really.

6589   MR. CONDON: Commissioner, the technical answer is -- and I think an example that somebody gave yesterday was if they wanted an ultra low latency gaming solution. I can't remember who said that, but that's a good example of what Juniper meant by a whitelist. You know, they might signal that they want to start gaming with this.

6590   It is pretty complex, Commissioner. It is not widely deployed, and I am always very cautious about things like that. On a network of our size, we inevitably have a really good track record of breaking things because they don't scale.

6591   So I would be very cautious about the ability to deploy that in the short term. It's an interesting model, I agree, but I would like to see it go further in the market first.

6592   MR. DANIELS: We don't want to rule out anything. That is very clear. I think that a lot of people have come before you and said that who knows -- in terms of actually having the customer decide.

6593   I mean, whitelisting, to my understanding -- you know, what Primus is doing by choosing applications and giving extra -- and giving them priority over other applications is whitelisting in that sense, but getting down to the user level, where the user gets to choose, that may come.

6594   I wouldn't want to rule that out. The network would have to be created in a manner to be able to respect that.

6595   Obviously, that would be a good thing, it's a question of whether -- will it come? When it does, will it be attractively priced so that we can implement it? And, is that the best way to go? Perhaps.

6596   MR. CONDON: But we do have equipment of that nature in the labs, and we are studying it, yes.

6597   COMMISSIONER MOLNAR: Thank you.

6598   THE CHAIRPERSON: Suzanne...

6599   COMMISSIONER LAMARRE: Thank you, Mr. Chair.

6600   I want to try to answer one question that you asked this morning of us.

6601   At the start of your presentation you mentioned that, fundamentally, at the heart of this hearing is a dispute about innovation. Yes, but that's not the only thing that is at the heart of this hearing.

6602   A couple of times you have inquired why traffic management practices are an issue, since the vast majority of users are happy with the service they get. I get that point, but at a certain level we have to be concerned about all users.

6603   There comes a time when a certain set of users falls into the minority, even though they fall within the vast majority of users at times.

6604   That's why we are inquiring and being, maybe, picky, if you find it picky.

6605   Moving along, another comment that you made in your presentation was about the fact that people have made allegations that there are problems, that there are issues, but that they don't have proof. They don't have data to support the allegations.

6606   The problem probably is that that information lies with ISPs, it doesn't lie with customers.

6607   On top of that, yourself, you mentioned that your customers are not that interested -- at least that is your perception -- in all of the details of how the traffic is being managed.

6608   Is there any way, from your perspective, that you can demonstrate that, indeed, what you are saying, that there is no problem, that there isn't any problem?

6609   MR. VANDENENGEL: Us being Bell, when we do something wrong for the customers, we hear about it pretty quickly, and we hear about it in a number of ways.

6610   We do churn research. Customers say they would leave us if we had problems.

6611   We do research on acquisition and what customers are looking for.

6612   None of that data has pointed to peer-to-peer as an issue, our shaping of peer-to-peer file sharing as an issue. Really, the top of mind for customers are things like price, customer experience, and when there is a problem, that we get the problem resolved quickly.

6613   So nothing that we have points to peer-to-peer file sharing shaping as an issue.

6614   COMMISSIONER LAMARRE: Okay, so then let's talk about consumer experience and price.

6615   The Chair this morning mentioned that the advantage of being last is that you have heard everybody, you have heard our questions, and that, therefore, you can tailor your intervention. But the flip side of the coin is that we also do comparisons, so some of my questions are within those lines.

6616   When you are talking about consumer experience, one thing that I still do not get is the fact that you are throttling peer-to-peer both on the upload and on the download, and I am going to tell you why I still don't get it.

6617   Rogers throttles only the upload, not the download, and when Rogers explained why they are doing that, they said it is because their customers' experience is important as far as the download is concerned.

6618   And I pointed out the fact that what they brought forward was anecdotal, not scientific.

6619   And they realized that the upload was going to non-Rogers clients. So they were not so concerned about the upload, because whoever was getting upset or frustrated by it was not their client, but on the download, it was indeed their client.

6620   I get your point: Rogers is cable and you are DSL. Well, TELUS is DSL also, and TELUS does not throttle at all.

6621   Can you explain to me why you chose to still throttle the download?

6622   As far as I can tell right now, it doesn't seem to me to be a good, customer-tailored approach.

6623   MR. DANIELS: I can't speak to why TELUS doesn't. The only thing I can say is that, five years go, or even three years ago, if you would have had this proceeding, you would have been able to look at Rogers and say: Tell me why Bell doesn't and you do, because --

6624   COMMISSIONER LAMARRE: That is why I am asking you, why did you choose to still throttle the download?

6625   MR. DANIELS: Our problem was in the download. We faced that in the Aggregation Network, and it was a question of what was the best solution.

6626   We had spent in 2007 an additional $110 million in augmenting our network that was unexpected. We saw the problem continuing to grow, and around the world -- we are not unique in this -- we looked at it and determined that -- we looked at the DPI and said that if we have a problem, it is being caused, not primarily, not exclusively, but largely by P2P and what happens in a congested state, and we determined that this was, for our users, the overall better experience to do that.

6627   There are different characteristics that may exist in TELUS' network from ours, in terms of TELUS' customer base. I can't speak to what exactly has gone on with TELUS, but I can speak to -- and I am going to get my colleague to speak to it, because, as it is clear from the record of this proceeding, we don't in the Atlantic provinces, but we do in the central provinces, and there are reasons for that, and perhaps that can explain --

6628   I can't speak to TELUS, but we can talk about our own network, why we do and why we don't.

6629   Denis...

6630   MR. HENRY: As Jonathan says, in the Atlantic provinces we do not traffic shape. In the Ontario and Quebec network, it is integrated with the Bell network, and we have experienced the congestion problems there, and the network is managed the same as Bell's.

6631   In the Atlantic provinces, we have not experienced the level of congestion that Bell has. There are a couple of reasons for that. We have looked at per subscriber usage in the peak periods. We don't have the same level of peak period subscriber usage as Bell does.

6632   We have also looked at the topology of the network. Population density in the Atlantic provinces is less than in Ontario and Quebec. As a result of that, it allows the network to be designed in a way that there is less stress on the Aggregation Network.

6633   So we have not experienced the level of congestion; therefore, we do not employ traffic shaping.

6634   That doesn't mean that we won't get there. I don't know what the future will bring. We monitor the network. If we do hit the levels of congestion, we certainly want to have traffic shaping as a tool in our toolbox.

6635   COMMISSIONER LAMARRE: Okay. Thank you.

6636   On top of throttling the traffic, there are also economic measures that have been put in place by Bell recently. You have, actually, reduced quite significantly the monthly bandwidth cap of your least expensive package. Actually, you have reduced it by 10 times, from 20 gigabytes per month to 2 gigabytes per month.

6637   And if I am making a mistake, you can correct me, but I am just stating your numbers.

6638   Also, one of your most popular products went from 60 gigabytes to 25 gigabytes per month, and yet the prices, from what I see, are almost the same as they were before, if not the same.

6639   Wouldn't that be enough of a measure, and would you not need, on top of that, to throttle --

6640   I get the point about peak moments of the day, or of the week, or whatever, but by offering different packages like this, aren't you addressing most of the problem?

6641   MR. DANIELS: I am going to hand it over to my colleague Peter to talk about it in more detail, but I think, again, we have a three-pronged solution. We do not believe that pricing today -- to get to the peak, you have to get to per-minute pricing. Users have to understand the difference. For us, in our experience, it is not good enough simply to implement overall caps in the course of the month. If it were, we would -- if that change resulted in a complete change in our network so that we didn't see congestion, we would stop traffic shaping. We are not interested in doing it if we don't need to do it, which is why we limited it to the times and period of day that we have.

6642   I am going to let Peter answer that, but the one aspect that I would note is, also, right now, we don't even have -- I have to put this plug in -- we don't even have this usage-based billing, even at our old levels, for wholesale.

6643   You know that we have an application before you for that, but we don't even have that yet, so...

6644   Peter...

6645   MR. VANDENENGEL: I don't know what else I can add to what Jonathan just said.

6646   COMMISSIONER LAMARRE: Well, then, don't.

--- Laughter

6647   THE CHAIRPERSON: We are running out of time.

6648   MR. VANDENENGEL: For the record, I have to show that, in Ontario, we did lower prices. So that was a good thing, as well, for the consumer.

6649   COMMISSIONER LAMARRE: Okay. A different topic -- wireless. In your submission this morning, do I understand correctly that the guidelines that you are proposing would be adequate to evaluate traffic management practices on wireless -- mobile wireless?

6650   MR. DANIELS: Absolutely.

6651   In fact, what we state in the guidelines, if you look at that one page where we sort of summarize, we say, specifically, in limiting the negative impacts -- because I don't think there is an issue between transparency and privacy.

6652   There is no need to make distinctions between a wireless network and a wireline network but in limiting the negative impacts, we say in the second line:

"Implicit in the determination of what is 'reasonable' is the recognition that different networks face different problems and therefore ISPs require the flexibility to reflect those distinctions in adopting ITMPs."

6653   So I guess what we are trying to -- and then we give an example specifically at b. about wireless versus wireline, very explicit in there.

6654   The reason we put that language in is ultimately some people can say, you need a completely different test. We think the test is reasonableness.

6655   But, of course, when you look at whether you are reasonably limiting the negative impacts, what you do in a wireless network may be quite different than what you can do in a wireline to achieve that objective, just in the same way that us impacting the downstream on a DSL network is different from what goes on in a cable network where they are focused on the upstream.

6656   COMMISSIONER LAMARRE: And would you also agree that we need to make the distinction between mobile wireless and fixed wireless?

6657   MR. DANIELS: Potentially if that is brought forward to you at the time, if it is relevant about the nature of the distinction but we think we are flexible enough that if it is justified, it is part of the test.


6659   Je ne peux évidemment pas vous laisser partir sans parler de l'impact des techniques sur la vie privée, et, Maître Morin, je dois vous dire que je me sens obligée de revenir sur la réponse que vous avez faite au président.

6660   Bon, je comprends que vous n'avez pas l'intention -- et c'est très, très clair là -- vous n'avez pas l'intention d'utiliser les données qui sont colligées dans le cadre d'inspection avancée des paquets pour d'autres fins que celle de la gestion du réseau.

6661   Ceci dit, vous avez aussi mentionné que, selon vous, c'était la loi -- on va l'appeler, on va prendre son acronyme anglais, qui est très commode -- le PEPIDA, qui faisait en sorte que, de toute façon, c'est PEPIDA qui vous empêchait d'utiliser ces données-là.

6662   Votre réponse ne me satisfait pas totalement parce que je la trouve incomplète, et je vais vous expliquer pourquoi, puis vous me direz si vous êtes d'accord ou non.

6663   D'un côté, PIPEDA est une loi de nature généraliste qui ne s'applique pas seulement aux fournisseurs de services Internet, et lorsque cette loi-là a été adoptée, il n'y a jamais personne qui a parlé de gestion de trafic Internet. Donc, d'essayer d'amalgamer les informations de trafic Internet sous PIPEDA, c'est une bonne idée, mais ce n'est peut-être pas la seule bonne idée.

6664   La commissaire à la Protection de la vie privée, dans sa soumission, a elle-même fait valoir que, et je vous cite sa soumission :

" ...que les rôles du CRTC et du CPVP prévus par la loi se complètent en ce qui a trait à la vie privée. Le CRTC a compétence sur les questions touchant la vie privée qui découlent des activités des réseaux de télécommunications. "Tel que lu)

6665   Donc, la gestion du trafic Internet, ça découle de vos activités de réseau. Donc, est-ce que vous êtes d'accord qu'on a aussi, sous la Loi des télécommunications, le devoir de nous assurer que ce qui est fait contribue à la protection de la vie privée?

6666   MME MORIN : Plusieurs choses. Absolument. Merci encore une fois de revenir. On est complètement d'accord que la Commission continue d'avoir un mandat pour la protection de la vie privée des abonnés.

6667   Cependant, comme fournisseur -- et puis tous les autres fournisseurs sont venus ici devant vous la semaine passée ainsi qu'hier -- nous sommes aussi assujettis à PIPEDA, comme vous l'avez bien expliqué.

6668   Lorsque la commissaire vous a écrit pour vous indiquer assurez-vous bien de regarder les implications du DPI ou bien de ITMP sur la protection de la vie privée, on est complètement d'accord, on a tous des... moi, je suis chez Bell Canada depuis 10 ans, mais on a tous des gens dans les organisations, les fournisseurs, que c'est notre responsabilité de s'assurer que l'organisation peut atteindre les obligations qu'on a en vertu de PIPEDA ainsi qu'en vertu de n'importe quelle règle qui s'applique, qui émane de la Commission.

6669   Alors, absolument, nous pensons que vous avez un mandat qui continue. Mais, de plus, ça fait des années qu'on se trouve en vertu de PIPEDA. On reçoit des plaintes de nos clients. Je ne peux pas me souvenir la dernière fois qu'on a reçu une plainte émanant de la Commission en vertu de la protection de la vie privée. C'est toujours des plaintes qui nous viennent par le biais de l'OPC, si je peux dire.

6670   Alors, on est très habitué à faire ça, et puis, même maintenant, comme vous le savez, nous avons une plainte que CIPPIC a déposée. Nous n'avons aucun abonné qui s'est plaint, mais nous avons une plainte de CIPPIC devant l'OPC, et nous sommes en collaboration avec elle et avec son bureau afin de régler cette plainte, et ça se peut que peut-être on va mettre d'autres informations sur notre site web.

6671   Mais, en général, elle est très, très bien placée pour regarder n'importe quelle de ces questions. Il n'y a rien d'unique dans notre secteur. Cette même Loi, PIPEDA, s'applique dans le secteur de l'assurance, dans le secteur bancaire, et tous, comme organisations canadiennes, on suit toutes les mêmes règles.

6672   CONSEILLERE LAMARRE : Mais on est peut-être mieux placé, ici au Conseil, pour évaluer à quel point certains techniques peuvent être plus invasives que d'autres, de gestion de trafic Internet.

6673   Ça, ça m'amène à la question du shallow inspection plutôt que le deep packet inspection. Comme le président l'a mentionné, la question qui avait été soulevée par le professeur Coates relevait de la protection de la vie privée.

6674   Alors, est-ce que vous êtes en mesure, vous, de nous dire, selon vous, laquelle des deux techniques est plus ou moins invasive? Est-ce que, avec le DPI, vous obtenez des informations qui sont plus détaillées sur l'utilisation des applications par les usagers que si c'était seulement du shallow inspection?

6675   MME MORIN : Il se peut que je vais demander à mon collègue Carl, mais juste...

--- Off-the-record discussion

6676   MME MORIN : Mais afin de juste toucher la question de la vie privée encore une fois, nous ne retenons aucune information reliée au paquet auquel s'applique notre technologie. Alors, cette information-là, ce n'est pas quelque chose qu'on garde.

6677   CONSEILLERE LAMARRE : Oui, je suis d'accord, sauf qu'il demeure que vous la surveillez cette information-là. Alors, vous avez beau ne pas la conserver, vous la surveillez, et ça aussi, c'est une question de vie privée.

6678   Si vous venez voir l'agencement de mes meubles dans mon salon mais que vous ne retenez pas comment c'était mal agencé, il n'empêche pas que vous êtes venue voir dans mon salon. Alors, je pense que c'est à ce niveau-là que parfois, il y a une inquiétude et aussi un agacement de la part des utilisateurs.

6679   Alors, peut-être que oui, votre collègue peut nous donner une réponse.

6680   MME MORIN : Juste avant, pour que Carl puisse savoir aussi, ça se fait machine à machine, hein! Alors, ce n'est pas quelqu'un qui est assis là qui regarde. Ça se fait par la machine. On applique la politique nécessaire pour traffic-shaping, puis après ça, c'est fini, c'est terminé.

6681   Alors, je vais demander à Carl, et puis, je vais le répéter peut-être en anglais pour lui pour qu'il puisse complètement savoir.

6682   LE PRÉSIDENT : On a des traductions simultanées.

6683   MME MORIN : O.K.

6684   LE PRÉSIDENT : Nous sommes vraiment pressés pour le temps.

6685   MME MORIN : O.K.

6686   MR. CONDON: (Off microphone).

6687   THE SECRETARY: Your microphone, please.

6688   MR. CONDON: If I understood your question correctly, Commissioner, you are asking is there a privacy issue.

6689   I think the intent of -- I know the intent of this shallow packet inspection is identical in that sense to deep packet inspection. It is to --

6690   COMMISSIONER LAMARRE: I know the intent is the same but what I am trying to get at is with deep packet inspection, do you get more detailed information about what the user is doing with the application than you would with shallow packet inspection?

6691   MR. CONDON: I don't --

6692   COMMISSIONER LAMARRE: Yes or no or you don't know.

6693   MR. CONDON: I am just thinking that through. You could, yes, Commissioner. You would -- there is a little more information available.


6695   MR. CONDON: However, I would caution to Rogers' points yesterday, it is not as accurate.

6696   COMMISSIONER LAMARRE: Okay. I did get that point, so thank you for repeating it though.

6697   You also mentioned earlier in answer to the Chairman that to your knowledge there is no commercial equipment available to do shallow packet inspection instead of deep packet inspection.

6698   Now, there are two sides of a market: there is the demand and there is the offer. So if nobody ever asks, obviously nobody -- it may be difficult at one point to get the product on the market.

6699   Has Bell ever put forward or can Bell start doing testing with shallow packet inspection? Because you did make that statement, it is not as accurate. Well, if you don't have the equipment, you haven't tested it, you cannot make the statement that it is not as accurate until you actually test it.

6700   MR. CONDON: I would disagree with you on that point, Commissioner.


6702   MR. CONDON: I think we understand enough about the protocols and their behaviour and the masquerading to state with certainty that it wouldn't be as accurate.

6703   MR. DANIELS: I think that one other aspect that I know, because we did have a discussion with our vendor about this and one thing to point out is that they use flow today -- as I think was described, it is part of the solution. First, you look in the packet and then you also look for the flow in terms of that aspect.

6704   But if that was the sole measure, it is very easy for P2P protocols to change the characteristics of the flow. It is easier to mask that aspect than it is to do the encryption that they are doing today.

6705   COMMISSIONER LAMARRE: So you would not meet your objective of trying to manage the congestion at peak hours on the network, that is what you are saying?

6706   MR. CONDON: Yes. And I would point out that it is also much more resource-intensive and it is a very much more expensive thing to do.

6707   COMMISSIONER LAMARRE: Going back to your guidelines which you are proposing in lieu of an Oakes-type test, you do mention that:

"The Oakes Test is a constitutional test that is fundamentally about limiting government intervention into private affairs."

6708   I would state it a little bit differently, saying it is a constitutional test to ensure that fundamental rights of Canadian citizens are protected.

6709   Now, privacy being a fundamental right, if we had to evaluate a traffic-management practice which we found infringed or violated privacy, don't you think that we should in such cases adopt an Oakes-type test in order to evaluate the relevance of that practice?

6710   MR. DANIELS: I guess I have a couple of things to say on that.

6711   Number one, I think it is really a very big difference to talk about the government as opposed to a private party, in terms of the Constitution gives fundamental rights to people that can't be infringed by the government, right.

6712   COMMISSIONER LAMARRE: I agree with you but at the same time --

6713   MR. DANIELS: So that is --

6714   COMMISSIONER LAMARRE: Just let me finish.

6715   MR. DANIELS: Yes.

6716   COMMISSIONER LAMARRE: At the same time, section 7(i) of the Telecom Act requires that whatever we choose to do we do it in a way that contributes to protection of privacy and we are a branch of the government, if you may phrase it that way. As an administrative tribunal under federal jurisdiction, we have the obligation to also protect and respect Canadians' fundamental rights. So that is where I am coming from. You may go on.

6717   MR. DANIELS: I don't dispute anything you said but I think that there is a fundamental difference between saying, if it is the government interfering in -- the government and you being a branch of the government, in terms of how you go about making the rules that you do as opposed to taking and saying, by the way, we want all the private parties to be treated as if they are the government and subject to the same standard and test as if they are acting as the government in terms of what they do.

6718   We are not the government. We are not subject to the same test. People have different options. They can choose different providers if they are not satisfied. It is not top of mind. It is not -- we are telling you that for the vast majority of customers it is not an issue.

6719   But I hear you say: Ah! But we are here to worry about also -- not exclusively the top priority -- we also have to worry about the minority of users that are impacted.

6720   And I agree and I support that notion and I come back and say: So what is the appropriate test in that situation?

6721   And we have come back and we have proposed to you and said: The statute requires reasonableness. And that is my fundamental problem, is that we are taking a test of constitution about when the government, the ultimate monopoly, is coming in and interfering in individual rights on that aspect. They are coming in and you are saying: Well, hold on, we want to subject you to the same test and the same aspects and we don't think that is appropriate.

6722   COMMISSIONER LAMARRE: Not even to protect fundamental rights of privacy?

6723   MR. DANIELS: I am being careful. I mean, privacy -- I don't want to downplay the importance of privacy. We believe in privacy and privacy for our customers.

6724   But I will also note, as my colleague explained to you, that under privacy law in Canada there are all sorts of companies that in private enterprise take all sorts of private information into account and there are rules around it, as long as they have customer disclosure, they are allowed to use it.

6725   People can use -- you know, when I am doing a search on Google, I know what I am calling up in terms of that and Google uses it to also target ads at the same time. Is that an invasion of privacy? I mean it is understood and consented by the customer.

--- Laughter

6726   MR. DANIELS: But that is what I am trying to say, it is not the same. It is not the same level.

6727   COMMISSIONER LAMARRE: I get your point. I get your point.

6728   Lastly, I just want to point to something you -- it is actually a comment and if you wish you can respond to that in your final submission, which I am sure you will provide us with a final submission.

6729   In paragraph 20 of your submission this morning, your presentation, you quote what ARCH has said concerning the interpretation of section 27 and the Oakes Test.

6730   Well, I would like for you to take notice also of paragraphs 2349 to 2361 of the transcript, in which there was a discussion between ARCH and our Chairman, in which the bulk of the discussion is not that ARCH was disputing the relevance of an Oakes test to evaluate traffic management practices but basically that they were concerned that if we did so, it would result in discrimination that otherwise would not occur.

6731   MR. DANIELS: I understand the distinction that they were making.


6733   MR. DANIELS: Having said that, the reason why I pulled it out is because fundamentally I think what they actually said here about the jurisprudence of section 27 is correct.


6735   MR. DANIELS: They are coming at it for a different reason. They are also -- you saw we have some dots in here. They had some comments about section 36 which I removed for flow purposes more than anything else -- pardon the pun -- when I came before you.

6736   So I appreciate that they had a different intention in mind. However, I think that their statement was bang on in terms of that one sentence about section 27.

6737   COMMISSIONER LAMARRE: Thank you. Those are all my questions.

6738   Merci, Monsieur le Président. Merci.


6740   COMMISSIONER DENTON: Lady and gentlemen, good morning, and my questions shall be brief.

6741   In paragraph 61 of your reply comments, you mentioned that you were exploring a "more granular approach" to bandwidth management.

6742   Can you explain or expand upon that statement, please?

6743   MR. DANIELS: I am sorry, what was the reference there?

6744   COMMISSIONER DENTON: In paragraph 61 of your reply comments, you mentioned that you were exploring a "more granular approach" to bandwidth management. Any thoughts or ideas on that?

6745   MR. DANIELS: Let me just pull it out to make sure that --

--- Pause

6746   MR. DANIELS: I think what we are -- like some of the ideas here are some of the things that people have put forward. Mr. Condon mentioned that we have some stuff in the lab in terms of some of the equipment that Juniper was talking about. We also are looking at the Cox solution, which is similar to what Primus has talked about.

6747   COMMISSIONER DENTON: The what solution?

6748   MR. DANIELS: The Cox solution. I am referring to Cox, which in the U.S. they are trialling in two areas, a distinction between time-sensitive and non-time-sensitive traffic rather than just strictly -- they do, I think, what Rogers said yesterday, was that Cox presently does exactly what Rogers does, which is focus on P2P.

6749   But Cox is also experimenting in two locations about having a time-sensitive versus non-time-sensitive distinction and that discussion is on the record, filed by -- I can tell you exactly in a moment but I think it was CIPPIC who filed -- or is it PIAC? I will tell you in a second.

6750   COMMISSIONER DENTON: Well, don't worry yourself.

6751   MR. DANIELS: Okay.

6752   COMMISSIONER DENTON: You are exploring about different and more precise means of traffic management based on different characteristics of the traffic; is that correct?

6753   MR. DANIELS: That is correct.

6754   COMMISSIONER DENTON: Would you like to add anything more, Mr. Condon, on that?

6755   MR. CONDON: No, I would not at this time.

6756   COMMISSIONER DENTON: I note that in the essentials plan you have dropped your bit cap by an order of magnitude. Was there any sort of basis for deciding on 2 gigabits per month? Is there some criterion for that?

6757   MR. VANDENENGEL: I mean, when we come up with products we look at the market. So we would look at our competitors, what the customers are wanting, and we also look at our management needs.

6758   I wasn't involved in discussions on the cap of 2 gigs but the thinking that goes through our heads or the larger group that goes through our heads is the competition, the pricing that the market will bear, and also our capacity and our network management activities.

6759   COMMISSIONER DENTON: Okay. That is a market-oriented answer. Thank you.

6760   Now, it may have been answered but I would like to hear it once again, if you haven't. I noticed that Primus uses traffic management practices during periods of congestion only. You use them all the time.

6761   Can I just hear one more time, if you haven't already stated, why you do it all the time?

6762   MR. DANIELS: Just to be clear, I think you mean why we do it everywhere. We don't do it all the time. We only do it from 4:30 to 2:00 a.m. But do you mean everywhere in our network? Because we don't do it all the time.

6763   COMMISSIONER DENTON: Okay. So some use different traffic management at peak hours and you are saying you use it only at peak hours, right --

6764   MR. DANIELS: That is correct.

6765   COMMISSIONER DENTON:  -- rather than just in periods of congestion only or you have determined that those are the periods of congestion?

6766   MR. DANIELS: Those correlate to our periods of congestion. That doesn't mean that we won't have a congestion period if there is a Michael Jackson funeral going on outside of that period but generally that corresponds to our period of congestion.

6767   And as I say, it is not pure 4:30 to 2:00 a.m., it is 4:30 to 6:00. We take -- P2P goes down to 512. And then from 6:00 to 1:00 it goes from 512 down to 256. So we even make a distinction within our peak period. We have, I don't know, maybe primetime peak as opposed to overall peak period. So we try to be granular in that one but we do it consistently throughout our network.

6768   COMMISSIONER DENTON: Now, it has been a basic contention of the retailers of your wholesale service that if more of the responsibility were devolved to them, coupled, I suppose, with improved signalling between the wholesaler and the retail, and, I suppose, coupled also with suitable payment for the costs, that traffic management at the retail level would be appropriate and sensible.

6769   I want to hear what you have to say about that.

6770   MR. DANIELS: Well, we certainly agree with them that traffic management at the retail level is appropriate and sensible. The question is: Is us instituting our traffic management practices on them when they buy the GAS service reasonable? And I posit to you that it is for a number of reasons.

6771   Number one, it is the shared network that is the same aspect. They have other options: HSA, unbundled loops, TPIA. We have heard that both TPIAs in Ontario and Quebec are not subject to it. So there are other options.

6772   But more fundamentally it comes down to the question: Is their traffic impacting ours when it is in the GAS? And we have demonstrated to you the statistics that show that it is.

6773   So it raises the question finally where Execulink says: But can I manage it? Says Execulink: I can manage it so it doesn't impact your traffic. Can you trust me to do that?

6774   Our problems are, number one, that we can't distinguish in our network between the various different wholesale providers. We can distinguish all wholesale but we can't make a distinction, Oh! Execulink is managing it correctly, someone else isn't.

6775   COMMISSIONER DENTON: Why is that? Could not the code be written whereby the origin could be distinguished? Is this just a software problem in terms of recognition of these sources of traffic?

6776   MR. CONDON: It is a protocol problem, Commissioner. The tunnelling that is used is dynamic.

6777   COMMISSIONER DENTON: Say that again. The tunnelling...?

6778   MR. CONDON: The tunnel IDs are dynamic. We can't tell the difference between them.

6779   COMMISSIONER DENTON: Tunnel IDs are dynamic and therefore they can be spoofed or hidden or just basically anonymized?

6780   MR. CONDON: No, that is not what I meant, Commissioner. I wasn't suggesting anything of that nature. I was just suggesting that at that level of the network they are not identified, they are not a permanent identifier to, say, Execulink.


6782   MR. CONDON: And it varies throughout the network.


6784   MR. DANIELS: There is one other aspect. So that is the technical issue but the other aspect is even if we could, right, how is it that Execulink would -- or let's take Primus.

6785   Let's say Primus has got -- I give priority over my traffic -- they have three levels of priority we have talked about, and at the bottom level, P2P and a bunch of other services are best-efforts. That is what they explained to you.

6786   How does that work when it comes to getting aggregated and shared in our network between all our retail customers and all other wholesale ISPs if they are not doing exactly the same shaping method that we are doing, exactly the same, they're trying to institute their own, it's going to conflict with ours in the sense of their P2P will still operate and have the impact on our retail traffic that we've talked about.

6787   It doesn't matter that they're giving priority -- extra priority in their network, even if they can pass it through ours for time-sensitive applications, their P2P will, unless we're shaping it, will have that impact that I talked about on our retail customers.

6788   The only way we could leave it up to them is if they instituted the exact same method that we do. And if we're going to do that, why don't we just do it for them which is what, you know, which is what we do, like, and not all of them can afford DPI and not all of them have DPI.

6789   I mean, as I told you, there's more than a hundred and some of them are very small.

6790   COMMISSIONER DENTON: It would seem to be then a matter of both of commercial and technical arrangements between you and the sellers of your wholesale service.

6791   MR. DANIELS: I think that's true, but I mean to think about it, it's not -- even -- like the only way to do it is that they'd all have to do exactly what we're doing, exactly the same manner in order so that it doesn't impact our retail customers and we'd have to make those arrangements -- those commercial and technical arrangements with a hundred -- north of a hundred ISPs that are on our network who purchase GAS service, which require them to go out and spend Cap-Ex to do what we're already doing for them.

6792   Like, I can't --

6793   COMMISSIONER DENTON: Well, if they wanted to spend the Cap-Ex, would they be prevented from doing so by your rules?

6794   MR. DANIELS: In terms of if they have their own DPI equipment?

6795   COMMISSIONER DENTON: Yes. If they wanted to go ahead and make the investments to conform to reasonable requirements, what would prevent them from doing so?

6796   MR. DANIELS: Nothing would prevent them, but the truth is we wouldn't -- today, as I say, every single one of them would have to do exactly that investment exactly the same way for it to be reliable, and what's the point of that to ask them all to do that when we're doing it for them?

6797   They can't -- this is the key, they're coming to you and saying, I want to be able to distinguish my traffic and have different ways of going about doing it better, equally reasonable.

6798   They can do that if they buy HSA, they can do that if they unbundle loop, but if they're on a shared GAS network where their traffic is going to impact our retail customers and being priced accordingly, then they can't have a different solution, they'd have to have the exact same solution as us and every single one of them would have to have that exact same solution as us because we can't distinguish between one ISP who does it and another one who says they're going to do it but they didn't.

6799   COMMISSIONER DENTON: Well, we'll watch that statement -- hold that statement for the future.

6800   Finally, another major contention that was advanced in this proceeding was basically the traffic management measures are a kind of byproduct of inadequacy of competition at the retail level.

6801   I invite you to comment.

6802   MR. DANIELS: Even in the models, without getting into too much detail that other people have pointed to, for example, people have pointed to England or the U.K. and said at that -- that, you know, that apparently is the panacea in terms of getting it right from terms of competition.

6803   Many, many ISPs, including BT and many ISPs there, other ISPs there institute traffic shaping, traffic shaping for P2P is found throughout the world on all sorts of levels of competition.

6804   Japan, which is sort of considered the leader in terms of building up broad band and fibre to the homes and so on, now my data's two years old, but in 2007 Japan had -- the industry did a survey and found that 35 -- 25 percent of ISPs there do traffic shaping and a further 11 percent of them are considering doing it, and that was two years ago.

6805   And I know two years ago we were considering doing it, we hadn't done it, so we'd fall into that category of the considering.

6806   So, I don't really think that to make the connection to say that the level of competition is responsible. It's the nature of the traffic and the economics in keeping broad band prices reasonable for the vast majority of users. Yes, we can have a rule that says you can't do any of this, that's going to ultimately I think impact customers and could drive up prices.

6807   I just don't see how that's in the public policy -- good public policy or in the interest of Canadians.

6808   So, now having said all of that, I know that people will be upset with me if I didn't also emphasize and, you know, we have another hearing about this, we think that it's highly competitive in Canada, very extensively competitive in Canada.

6809   We're changing our pricing, lowering our pricing, we're responding to the market and there's lots of ISPs out there that differentiate themselves. There's different solutions, there's options available.

6810   So, we don't agree with anyone who says that it's not competitive in Canada. We're going to have a hearing that's going to talk about that again later this year, so -- but I'm putting that aside for a moment in answering your question.

6811   I don't think that you're going to see a correlation between the state of competition and whether traffic shaping, traffic shaping's happening all over the world.


6813   MR. CONDON: Oh, I was just waiting for a supplemental, but in the U.K. BT Retail, Virgin, PlusNet, Sky, Orange and Tiscali all publicly disclose that they manage peer-to-peer traffic.

6814   COMMISSIONER DENTON: Thank you.

6815   That's my questions, Mr. Chairman.

6816   THE CHAIRPERSON: Thank you, Mr. Denton.

6817   Before I let you -- just reflecting on your answers while my colleagues questioned you and you're insisting on a reasonable standard as opposed to least intrusive.

6818   I think it's a bit of an appearance issue, for me. Least intrusive, you have to take into account not only the actual but also the reasonable -- as the effect. I mean, you do not only look in terms of what is the effect of it and, you know, you want to make sure that it is indeed a least (inaudible) but that it is not only on its appearance but in the effect and for all the various ISPs.

6819   And if you look at in terms of least intrusive in that expanded way, I think the difference between that and reasonable is relatively marginal.

6820   MR. DANIELS: Can I comment on that?


6822   MR. DANIELS: Because I think to directly address that, if I could get you to turn -- the best way to look at this is our one-page guidelines.

6823   THE CHAIRPERSON: M'hmm.

6824   MR. DANIELS: Because I think our goal is the same in that regard and I'm not going to repeat everything I said, but I think I've addressed it -- we've addressed it in terms of how we design.

6825   If you look, right before No. 1, we said:

" designing and implementing ITMPs to improve the overall experience for the vast majority of users."(As read)

6826   So, that goes to my point that I was discussing with Commissioner Lamarre about overall impact.

6827   However, Commissioner Lamarre says we have to be worried about the impact on the minority.

6828   Our definition of reasonableness says, ISPs should make reasonable efforts to limit the negative impacts of ITMPs on users, services, protocols and applications.

6829   So, when we say reasonable, we're talking about reasonable limits of the negative impacts which is directly designed to address the minority of users, or I mean, in our case, as I say, 50 percent of our users touch P2P at some point but only -- you know, 40 percent of them only use it four percent of the time. So, we're really talking about making sure that we're not impacting that 10 percent. Is that reasonable? And we're focused on the negative impacts, which is why we put that in there.

6830   So, I hope that would address the concern you have.

6831   THE CHAIRPERSON: Yes. As I say, I don't think we are very far apart, you're just coming at it from the other end.

6832   Thank you very much. We spent a bit more time with you than the others, but since you caused this in the first place...

--- Laughter

6833   THE CHAIRPERSON: Anyway, I appreciate very much the completeness of your answer and please take advantage of the time to fulfil the undertaking and answer some of the questions that you couldn't answer today.

6834   That's it for today.

6835   Madame la Secrétaire, over to you.

6836   THE SECRETARY: Thank you, Mr. Chairman.

6837   This concludes the items on the agenda of this hearing. The hearing is adjourned.

--- Whereupon the hearing concluded at 1120


____________________      ____________________

Johanne Morin         Jean Desaulniers

____________________      ____________________

Sue Villeneuve         Beverley Dillabough

____________________      ____________________

Monique Mahoney         Madeleine Matte