ARCHIVÉ - Transcription de l'audience
Cette page Web a été archivée dans le Web
L’information dont il est indiqué qu’elle est archivée est fournie à des fins de référence, de recherche ou de tenue de documents. Elle n’est pas assujettie aux normes Web du gouvernement du Canada et elle n’a pas été modifiée ou mise à jour depuis son archivage. Pour obtenir cette information dans un autre format, veuillez communiquer avec nous.
Offrir un contenu dans les deux langues officielles
Prière de noter que la Loi sur les langues officielles exige que toutes publications gouvernementales soient disponibles dans les deux langues officielles.
Afin de rencontrer certaines des exigences de cette loi, les procès-verbaux du Conseil seront dorénavant bilingues en ce qui a trait à la page couverture, la liste des membres et du personnel du CRTC participant à l'audience et la table des matières.
Toutefois, la publication susmentionnée est un compte rendu textuel des délibérations et, en tant que tel, est transcrite dans l'une ou l'autre des deux langues officielles, compte tenu de la langue utilisée par le participant à l'audience.
TRANSCRIPTION DES AUDIENCES DEVANT
LE CONSEIL DE LA RADIODIFFUSION
ET DES TÉLÉCOMMUNICATIONS CANADIENNES
Examen des pratiques de gestion du trafic Internet des fournisseurs de services Internet
Centre de conférences
140, Promenade du Portage
le 7 juillet 2009
Afin de rencontrer les exigences de la Loi sur les langues
officielles, les procès-verbaux pour le Conseil seront
bilingues en ce qui a trait à la page couverture, la liste des
membres et du personnel du CRTC participant à l'audience
publique ainsi que la table des matières.
Toutefois, la publication susmentionnée est un compte rendu
textuel des délibérations et, en tant que tel, est enregistrée
et transcrite dans l'une ou l'autre des deux langues
officielles, compte tenu de la langue utilisée par le
participant à l'audience publique.
Conseil de la radiodiffusion et des
Examen des pratiques de gestion du trafic Internet des fournisseurs de services Internet
Konrad von Finckenstein Président
Len Katz Conseiller
Suzanne Lamarre Conseillère
Candice Molnar Conseillère
Timothy Denton Conseiller
Sylvie Bouffard Secretaire
Regan Morris Conseiller juridique
Chris Seidl Gérants de l'audience
Centre de conférences
140, Promenade du Portage
le 7 juillet 2009
- iv -
TABLE DES MATIÈRES
PAGE / PARA
Open Internet Coalition 167 / 991
Zip.ca 226 / 1341
Coalition of Internet Service Providers 249 / 1493
Vaxination informatique 287 / 1684
Jason Roks 300 / 1745
--- L'audience reprend le mardi 7 juillet 2009 à 0903
985 THE CHAIRPERSON: Good morning. Bonjour.
986 Madame la Secrétaire, commençons.
987 LA SECRÉTAIRE: Merci, monsieur le Président.
988 We will begin Day 2 with Item 5 on the Agenda.
989 I would now invite the Open Internet Coalition to make its presentation. Appearing for the Open Internet Coalition is Jacob Glick.
990 Please introduce your colleagues and you will then have 15 minutes for your presentation.
991 MR. GLICK: Thank you very much. Thank you, Mr. Chair, and thank you to the Commission for providing us this opportunity to appear today.
992 My name is Jacob Glick, I am Google Canada's Policy Counsel. I am here today on behalf of the Open Internet Coalition.
993 The Open Internet Coalition represents consumers, grassroots organizations and businesses working in pursuit of a shared goal, keeping the Internet fast, open and accessible to everyone.
994 Members of the Coalition include companies like Google, Skype, eBay, Amazon, Ask.com, Sony Electronics, Ticketmaster, Evite, CitySearch and others. It also includes civil society organizations like Free Press, Public Knowledge, and New America Foundation.
995 I am joined today by the Executive Director of OIC, Markham Erickson. Markham is a founding partner of Holch & Erickson LLP where he represents clients before regulatory agencies, courts and the United States Congress. His practice typically involves engagement on complex issues relating to the Internet, new technologies and nascent industries. He is also an expert in Native American law and policy.
996 Markham helped to write and negotiate many of the federal laws that govern the e-commerce and use of the Internet in the United States. He represented the United States before the Organization for Economic Cooperation and Development on matters relating to speech and content regulation on the Internet.
997 Also on the panel is Robb Topolski, Chief Technologist for the Open Technology Initiative of the New America Foundation. Robb's career includes 15 years in the fields of software testing and quality assurance at Intel Corporation and Quarterdeck Office Systems, with a focus on developing networking products.
998 His 2008 report "NebuAd and Partner ISPs: Wiretapping, Forgery and Browser Hijacking" raised public awareness and helped launch Congressional investigations into large-scale ISP use of DPI for commercial eavesdropping.
999 He is also well known in the tech community, because in 2007 he proved that his ISP, Comcast, was using DPI to block peer-to-peer uploads, a discovery resulting in an FCC proceeding that ultimately led to an order ending the practice.
1000 The fourth member of OIC's panel today is Free Press General Counsel Marvin Ammoiri. Marvin was the lead attorney for Free Press on the case brought against Comcast by consumer groups before the U.S. Federal Communications Commission. He is a Professor of Telecommunications, cyber law and cyber warfare law at the University of Nebraska and a Term Member of the Council on Foreign Relations.
1001 Our purpose in appearing today is to assist the Commission in coming to its determination of what kinds of network management practices fall inside or outside the bounds of Canadian telecommunications law.
1002 During these brief opening remarks, we intend to make four simple points:
1003 One: Innovation and its consequential economic benefits are driven by the open Internet;
1004 Two: Practices that undermine the Internet's openness are bad for innovation;
1005 Three: Some traffic management is normal and okay;
1006 Four: The way to differentiate between acceptable ITMPs -- a mouthful of an acronym but I'm going to use it for the rest of the presentation -- from ones that undermine innovation is to employ the principled, light-touch test we propose for the interpretation of sections 27(2) and 36 of the Telecommunications Act.
1007 So first, point number one: Innovation is driven by the open Internet.
1008 By virtue of its "openness" the Internet is a platform on which anyone with a good idea and reasonable technical skill can develop and deploy tremendous innovations, easily reaching billions of people worldwide; what many people call "innovation without permission". Canada's global competitiveness depends on encouraging this innovation.
1009 The Commission will be well aware of how the open Internet is a platform for innovation. After all, we just concluded the New Media proceeding in which many of the same Carrier ISPs that appear in this proceeding came before the Commission and extolled the virtues of the open Internet.
1010 In their testimony they told you how important the open Internet is to consumers. They told you that the world of new media content is about consumer choice; pull, not push. They told you how they are in the business of responding to consumers, not building walled gardens.
1011 Yet in this proceeding many of those same carrier ISPs ask you to legitimize the arbitrary restrictions they place on consumer choice by managing network traffic on an application-specific basis.
1012 We agree with the sentiments that the ISPs expressed in the New Media proceeding. The open Internet drives innovation in new applications, consumer choice and markets for Canadian producers.
1013 For a deeper economic analysis of how this works and a summary of the relevant literature, I refer you to Professor Hogendorn's evidence about the value of general purpose technologies like the Internet, submitted as part of our initial comments.
1014 As a matter of public policy, it is better to have more, robust access to the open public Internet. That's because the Internet is more than a service provided by carriers; it is an economic engine. It is indispensable infrastructure for our economy, our society and our democracy.
1015 As a regulator charged with promoting the public interest and the development of Canada's telecommunications infrastructure, we submit that the Commission ought to regulate in a manner that promotes the development of the open Internet.
1016 The second point, which is a corollary to the point that the open Internet drives innovation, is that practices that undermine the Internet's openness are bad for innovation; in this case, application-specific traffic management practices. ITMPs that discriminate between applications distort market forces and harm user choice.
1017 Very simply, when one particular application is slower than others it is less attractive to users. Giving carriers the power to slow down applications at their own discretion will change user behaviours, distort innovation and undermine the competitive market in applications. This point is undisputed in the evidence before you, both in this proceeding and in the New Media proceeding.
1018 Third, some traffic management is normal and okay.
1019 Congestion is not new on the Internet, but the current hysteria about traffic growth should be kept in perspective. As the Internet has grown there have been various times when the network has be strained. After all, it wasn't that long ago we started with an Internet that exclusively transmitted text. With the advent of the World Wide Web and the commercialization of the Internet, it became more commonplace to see photos, then sounds and short videos online. All of these multimedia formats strained our existing network at the time.
1020 Similarly, as orders of magnitude, more people got online and the network experienced various kinds of congestion. Throughout this evolution, the Internet has seen greater strain than it sees today.
1021 Increased capacity has been the primary means of dealing with this evolution. Imagine if 10 years ago network operators had managed for scarcity instead of creating abundance. Think of the application-specific throttling that would have happened against HTTP in 1993 and photo-sharing applications in the mid 1990s.
1022 The Commission should be careful not to adopt policies that encourage scarcity and dis-incent -- I know dis-incent isn't a real word, but we all know what we mean -- dis-incent network operators' investment in adding capacity.
1023 Having said that, the point we make in our written submissions, and the point we want to emphasize today, is that there is nothing, per se, wrong with traffic management. Some traffic management practices are fine, and some are problematic.
1024 As we will explore later, OIC proposes a principle-based approach for distinguishing between useful and neutral ITMPs from ones that undermine openness and distort innovation.
1025 We urge you to reject as false the choice between debilitating network congestion and application-based discrimination. This is a false dichotomy. The evidence is, and the experience in Canada and the U.S. already shows, that carriers can manage their networks, reduce congestion, and protect the open internet all at the same time.
1026 The fourth point: OIC's three-step test.
1027 We urge the Commission to adopt a principle-based, light-touch regulation that includes flexibility for carriers and solves congestion, while preserving openness. This principled approach need not be invented from scratch. We propose an approach rooted firmly in sections 27(2) and 36 of the Telecommunications Act. The test is designed to answer the questions posed by 27(2), what constitutes unjust discrimination, and posed by section 36, when should the Commission permit carriers to interfere with the purpose of a communication in the context of traffic management.
1028 As an aside, we note that it seems obvious to us that discrimination between applications is discrimination for the purposes of 27(2). If you require further convincing on this point beyond what is in our initial comments -- we have, I think, an in-depth exploration of this point in our initial comments -- we would be delighted to take questions on that point.
1029 With respect to 27(2) and 36, we submit the following three-step test for your consideration.
1030 One: Does the traffic management practice in question further a pressing and substantial objective?
1031 Two: Is the traffic management practice narrowly tailored to address this objective?
1032 Three: Is the traffic management practice the least restrictive means to reach the objective?
1033 Step one: Does the traffic management practice in question further a pressing and substantial objective?
1034 Relieving specific instances of network congestion may constitute a pressing and substantial objective. However, the evidence in this proceeding does not suggest debilitating network congestion, nor one particular cause of congestion. More accurately, the cause of congestion felt by some carrier ISPs is the growth of high bandwidth content, particularly video.
1035 Some of these carrier ISPs experience this as an increase in the use of some peer-to-peer applications. However, that is simply because those applications are widely used to distribute high bandwidth content.
1036 Again, as an aside, there is this misleading charge that BitTorrent and other P2P applications exploit -- and I put that in air quotes -- the internet. As the person who discovered Comcast's BitTorrent Shenanigans, Rob can answer all of your questions on this topic.
1037 Step two: Is the traffic management practice narrowly tailored to address the objective?
1038 Carrier ISPs need to show that what they are doing actually solves the problem they are trying to address, without being over-inclusive or under-inclusive, without causing other unintended consequences.
1039 Throttling particular applications will almost never be narrowly tailored. It will almost certainly be both overbroad and underbroad -- overbroad because it throttles uses and users who are not causing congestion, like the user who might use BitTorrent occasionally to download a new version of Linux, and underbroad because other high bandwidth uses will not be captured, like the user who updates her Windows OS during peak times by downloading a larger service pack.
1040 Moreover, as discussed already, and detailed in our written submissions, application-based traffic management techniques have all sorts of negative effects on innovation. They undermine competition and innovation, without permission, at the heart of the internet.
1041 Even if they work in alleviating some congestion, they are bad for innovation.
1042 And, in any event, there are other, better means of lessening congestion, which is where the third branch of the test comes in: Is the traffic management practice the least restrictive means to meet the objective?
1043 In our comments, we indicate a number of other traffic management techniques that do not have negative consequences for innovation, and do not discriminate between applications. These include, first and foremost, adding network capacity.
1044 They also include a number of application-neutral and price-based levers.
1045 Carriers should have to prove that these techniques do not work before they are authorized to discriminate between applications.
1046 OIC's proposed test is similar to the one developed by the FCC to handle similar traffic management questions. It is also rooted firmly in Canadian law, drawing its inspiration from the Charter analysis that has developed around section 1.
1047 Our test can be used to determine whether the particular traffic management technique is unjust under section 27(2), or whether it is interference with the meaning or purpose of a telecommunication under section 36.
1048 This is a technologically neutral test, one that can be applied in a multitude of circumstances, and one that does not put the Commission in the place of a network engineer.
1049 In sum, we have four simple points:
1050 One, innovation is driven by the open internet.
1051 Two, practices that undermine the internet's openness are bad for innovation.
1052 Three, some traffic management is normal and okay.
1053 And four, the way to differentiate between acceptable ITMPs from ones that undermine innovation is to employ the principled, light-touch test that we propose for the interpretations of sections 27(2) and 36 of the Act.
1054 With that in mind, we urge the Commission to adopt the test we propose and make a finding that application-specific ITMPs are contrary to sections 27(2) and 36 of the Telecommunications Act. In so doing, the Commission can provide an immense amount of flexibility for ISPs, while at the same time keeping them from undermining the core openness of the internet.
1055 Those are our comments. We look forward to your questions. Thank you, merci.
1056 THE CHAIRPERSON: Thank you very much for your presentation.
1057 I am delighted to see you here. As I told you, I missed you at the New Media Hearing.
1058 Google, as one of the large players on the internet, should have been there, and I am glad to see that you are here today.
1059 Looking at your submission, you base everything on the issue of the openness of the internet. How exactly would you define "openness" here?
1060 Let's make sure that we understand what is "openness".
1061 Obviously, you have to be an internet subscriber, you have to pay a fee, et cetera. So what exactly does "openness" mean to you?
1062 MR. GLICK: I am going to ask Markham Erickson, as the Executive Director of OIC, and someone whose title has "openness" in it, to field that question.
1063 MR. ERICKSON: Based on that, I should be able to answer that question, I hope.
1064 Thank you for the question.
1065 "Openness" is essentially what Jacob referred to in the opening comments, which is an internet that allows for innovation without permission. It creates an environment where the essential infrastructure of communication can be used both by users, to engage in speech with the worldwide community without having to get permission to engage in such speech, and it allows application providers to have certainty to know that they can develop new technologies and introduce those technologies over the internet, without having to go through a gatekeeper or ask permission to introduce those technologies to the internet, which allows for users to make decisions about which applications will succeed and which applications will fail, rather than have a gatekeeper make those decisions before those applications are able to reach the worldwide audience.
1066 THE CHAIRPERSON: So innovation without permission, subject to the normal constraints of law, presumably.
1067 MR. ERICKSON: Yes, that's right.
1068 THE CHAIRPERSON: Now, your test that you propose is essentially based on the Charter. We call it the Oakes Test in Canada. It is basically a variation of it, except that you left out the proportionality point.
1069 As you know, I am a former judge, and I had to apply this test in the rights context, and it is a very nice analytical framework, step-to-step, but at the end of the day you are forced, in the human rights context, to make, essentially, a value judgment, a value judgment as to whether something is or is not least intrusive, whether it is pressing or substantial.
1070 What is pressing and substantial? After you go through all sorts of stages of analysis, at the end of the day, you make a judgment.
1071 Here we are dealing with industry, we are dealing with commerce, et cetera. Can we quantify those, so that it becomes an objective test rather than a subjective test?
1072 Because that seems to me to be very important.
1073 How would you measure what is pressing or substantial, for instance?
1074 MR. GLICK: I am going to ask Marvin to answer that, because this test was largely something that he helped draft in the U.S. context.
1075 MR. AMMORI: I think that one of the strengths of the test is that it is flexible and you can apply it as new circumstances come up, and you can build precedent.
1076 Essentially, what you have is what you referred to yesterday as a guideline, and then you can apply it to the actual facts at hand. You can have experts like Rob and experts from the industry discuss the technical matters, and you can have economists discuss the importance of keeping the network management tools as general and as evolvable as possible.
1077 One of the strengths of it is that it can be applied on a fact-by-fact basis, with guidelines for investment as the engine in the centre.
1078 I think there is no way to avoid some sort of value judgment, because what we have here is basic infrastructure for our economy, and we have to ask, how are we going to regulate and ensure and protect everyone who uses this basic infrastructure.
1079 So there is no way out of the conundrum of making it purely objective, there will be some subjectivity involved.
1080 THE CHAIRPERSON: You can introduce objective elements, but at the end of the day whether it is pressing or not is a value judgment.
1081 MR. AMMORI: Yes, and one objective element, or pretty objective element, is the application-specific nature of throttling or of the network management.
1082 So if it is application-specific, if it is focused on BitTorrent or VoIP or something else, then there is a higher threshold and the test needs to be met.
1083 If it is application-neutral and if it gives lots of choice to the user as to how to prioritize and manage congestion in times of peak usage, that would be less restrictive, and so there are certain guidelines that are very helpful. But in the end there will be a value judgment made by this Commission after taking evidence.
1084 THE CHAIRPERSON: Just as a matter of interest, why did you not introduce any element of proportionality? For instance, CIPPIC, who we will be hearing after you, has put that in effect the action taken should be proportional to the evil that you are trying to fight.
1085 MR. AMMORI: I think we believe -- at least the way the U.S. test is structured, I believe -- that narrow tailoring includes proportionality, that it is not narrowly tailored to the problem if it is wildly overly burdensome. So the conclusion in the Comcast case was this is disproportionate to the problem, it affects all these different things.
1086 THE CHAIRPERSON: I see. Okay.
1088 COMMISSIONER DENTON: So I take it that if we devised a test, you would not object to proportionality being a part of that test?
1089 MR. AMMORI: No, not at all. I think that would be important.
1090 COMMISSIONER DENTON: Okay. I wanted to talk about your various steps of analysis here.
"Does the traffic management practice in question further a pressing and substantial objective?" (As read)
1091 First, I think, is who raises the -- who in your system or conjecture do you expect will be raising the issue? Do you expect a complaint-based system?
1092 MR. GLICK: I will pass this to Marvin in a second, but just to take a first gloss at it, it will be in some respects both ex ante and ex post. So we are expecting a set of ex ante guidelines issued by the Commission that will allow network operators to know more or less the rules of the game and to know, for example, that application-specific traffic management techniques are per se or prima facie offside the test and that they need to seek specific exemptions in order to have those authorized.
1093 But in other contexts where there is not a per se ex ante restriction, we would see ex post complaints coming to the Commission on a case-by-case basis, based on -- Marvin has an addition to that.
1094 MR. AMMORI: I just wanted to clarify that with an ex post complaint process which I think is flexible and helpful for the Commission, we would want to make sure that the carriers also engage in a lot of disclosure so that the public can monitor what the ISPs are doing and bring complaints based on disclosure as to the kinds of things that are being used.
1095 COMMISSIONER DENTON: You finished your question?
1096 MR. AMMORI: Yes.
1097 COMMISSIONER DENTON: Okay. I noticed in your February evidence that you gave a number of application-neutral traffic management practices that would be acceptable to you and I just want to ask you again -- I probably know the answer but I want to hear it again.
1098 Your basic concern, I take it, about traffic management practices which are not application-neutral is that they shift the balance of power from innovators to carriers; is that your basic concern?
1099 MR. GLICK: I don't know about the balance of power, but we think that the way the Internet has evolved by allowing application developers more or less to develop and deploy without seeking permission has worked really well.
1100 I don't know if I would put it in the context of power relationships but I think the answer is yes kind of.
1101 COMMISSIONER DENTON: Okay. I will take that for a yes.
1102 COMMISSIONER DENTON: Because basically if carriers were allowed to determine on an application-specific basis, then you are saying that that would start violating the principle that application developers must be able to innovate without permission and this would transfer the permission to carriers to determine which applications are suitable?
1103 MR. GLICK: Yes.
1104 COMMISSIONER DENTON: Okay. That is what I meant.
1105 MR. AMMORI: If I can just add. I represent a consumer group with 500,000 activists and we lead a coalition of about 2 million people and 800 organizations on net neutrality called Save the Internet in the U.S.
1106 So in terms of the balance of power, we are also concerned about consumers and citizens who want to be able to communicate without needing permission, without needing to -- you know, communicate however they want, through BitTorrent, through video. You know, there are open source video applications like Mirro where people have thousands of individual channels using BitTorrent in other applications.
1107 So we are also very concerned about the "balance of power" among citizens and people who control the bit structure.
1108 COMMISSIONER DENTON: So you envisage a communication system not merely as a set of economic actors but as a way for people to communicate among themselves?
1109 MR. AMMORI: A communication system, yes. People should be able to communicate amongst themselves.
1110 COMMISSIONER DENTON: A good idea.
1111 MR. AMMORI: Yes.
1112 COMMISSIONER DENTON: Now, price-based levers and other things, you mentioned, I think it was in passing, but I am interested in your view as to the relative balance or utility of open price-based discriminators to obviate the problem of traffic congestion.
1113 MR. GLICK: What I think we will see is a variety of techniques used, some of which use demand and pricing as levers to control how usage on the Internet evolves.
1114 So the basic claim, I think, underlying some of what the carriers are saying is that there is a negative externality being caused by use of particular applications, and they call this overuse, for what it's worth. And one of the ways to internalize a negative externality historically is through pricing mechanisms.
1115 So there are all sort of different kinds of pricing mechanisms we might see evolve and, in fact, I think that Canadian carriers are already deploying to varying degrees.
1116 COMMISSIONER DENTON: What about the transparency? Is it of concern to you that various price-based means of dealing with congestion -- does that have implications for either the innovation issue or the freedom to communicate issue to you guys?
1117 MR. GLICK: I think the answer is yes. You know, although we think that price-based levers are application neutral, they may have other unintended consequences in terms of retarding Internet use overall, discouraging Internet use. So all of us who have, you know, a mobile plan with a high data rate know what it's like to limit our use of the Internet.
1118 But ultimately, the Commission isn't in the business of regulating the fees that ISPs charge for retail Internet access and the role for the Commission in this context, I think, is to continue to promote competition at the retail broadband level.
1119 MR. AMMORI: Just if I can bring the American perspective for a moment. When the question of what we often refer to as metering, sort of charging extra per month for lots of bandwidth use, when that first came up, we as consumer groups said that could be acceptable under certain circumstances if it's not priced in what looks like an anti-competitive way to, say, target online video.
1120 There were trials by Time Warner Cable, which is the cable system in New York and in Texas, to try to charge per megabit based on certain equations, and the equations that were proposed in this trial seemed like they were targeted at online video. They seemed really problematic and the public rose up in anger. And there was a bill introduced in Congress by a Congressman in New York because his constituents were so angry.
1121 So one thing is that metering prices should still be subject to an ex post review. And in the U.S. it's unclear if it will be done by the anti-trust authorities or by our communications regulator. But we think metering should still be subject to an ex post review, though not an ex ante review because of the risk of anti-competitive pricing, because of the incentives that carriers have when they are video providers and there is online video, for example.
1122 COMMISSIONER DENTON: Thank you.
1123 Now, you have mentioned the question of building network capacity, and if I were in the shoes of a carrier, I am wondering whether that is just assuming the problem away. If I were they, I might respond, if you just propose that we just build network capacity, the network will be flooded with ever more traffic in any case. Such as building highways, they just get -- more cars can be produced.
1124 So how would you answer the concern that they might have that just proposing the increase of network capacity might be the optimal solution up until the point where the networks get flooded again?
1125 MR. GLICK: I am going to ask Markham to answer that, but I will add briefly that (1) the history of the Internet is one of more or less doing precisely that, building more network capacity to handle the increased demand, and that has actually worked really well and has allowed for the Internet we all like and all the services that we all use and enjoy.
1126 And (2) we are not, you know -- we are also, you know, not proposing that solely. We are also saying that some forms of traffic management are inevitable and acceptable. But I will let Markham elaborate on that.
1127 MR. ERICKSON: Sure. Thank you for that question. It is a very good question.
1128 What we have said in our comments is that -- is to point to the history of the Internet. Over the last 25 years there have been, over that time period, concerns that the Internet would be strained and historically the response to that has been to increase capacity rather than to introduce electronics or to try to manage the scarcity.
1129 So the answer historically has been to increase capacity, which has had the benefit of also spurring innovation. So there is definitely a virtuous circle of beneficial relationships between users, application providers and network provides, and hopefully there is a win-win for all three of those stakeholders. So as capacity increases, you see more innovative applications, you see more users subscribing and figuring out how to interact with each other in those ways.
1130 We cite too an Internet2 study from 2006. Internet2 is an organization that represents 400 leading universities and technologists and other institutions and they looked at this issue. They studied whether it was more efficient from both an economic and technological perspective to introduce quality of service electronics to manage the congestion, manage the scarcity or to increase the capacity.
1131 And that study determined that it was more efficient both economically and technologically to increase capacity rather than to introduce those electronics and manage scarcity, and there's a lot of other collateral issues that those electronics introduced.
1132 So we point to the past record as evidence that adding capacity is a more beneficial way for all participants in this ecosystem to deal with increased user activity, increased congestion.
1133 Having said that, we are not opposed to network operators managing congestion in other ways and we would give them a great deal of flexibility to do that. It is only when they engage in anti-competitive or an application-specific throttling or degradation where we think the Commission should have a clear rule that in an ex ante way that is going to be impermissible unless they can satisfy the test we've proposed.
1134 COMMISSIONER DENTON: Finally, the thing that strikes me -- and I ask you to respond -- is as we withdraw from price regulation of individual prices we find that there's another entire area of regulatory activity which is whether we're going to -- the test you have proposed.
1135 Do you think it reasonable to suppose that in the future that we're going to be -- the CRTC will be considering these issues of traffic management practices on a permanent basis, or do you see this as being something that will be a temporary phase of our work?
1136 MR. GLICK: I think it depends on the kinds of techniques employed by carriers to manage the networks.
1137 But I certainly think it's not unreasonable to think that after setting out clear guidelines for carriers that the landscape clears up nicely and that the Commission is not overly burdened by requests or, you know, similar type of issues.
1138 Markham wants to add something.
1139 MR. ERICKSON: I wonder if I could just add something to that, if that's okay, which is that, you know, I think you'll see and based on our past experience periods of time where there will be more congestion based on certain activities and applications or a period of time where we are in terms of network capacity where there's more tension.
1140 And the key I think is not to over react to those moments in time by giving carriers the incentive to manage scarcely rather than to invest in the network or to do things that would be -- that would cause collateral damage to that useful Internet ecosystem.
1141 But that over time we also have sort of Moore's Law working and you have technologies that allow for better compression rates and you have applications that figure out how to compress more and more content into smaller and smaller pipes as well.
1142 So, all of that is working together and in the ideal world that is able to work without -- with minimal regulatory interference. What we don't want to see is hopefully an over reaction to any particular moment where carriers are allowed to do things that would upset that ecosystem, that healthy ecosystem.
1143 COMMISSIONER DENTON: Thank you.
1144 Mr. Chairman, those are my questions.
1145 THE CHAIRPERSON: Thank you.
1147 COMMISSIONER KATZ: Thank you, Mr. Chairman.
1148 I've got a couple of questions. In your filing of February 23rd on page 6, paragraph 18, you state in the middle:
"Adding capacity to the network is an important public policy goal." (As read)
1149 COMMISSIONER KATZ: Can you elaborate on where you find that goal and is that a Canadian goal, an American goal, an international goal? Whose goal is this?
1150 MR. GLICK: Sir, I would think it's a Canadian public policy goal, it's also a U.S. public policy goal.
1151 Certainly the development of Canada's telecommunications infrastructure is a CRTC Telecommunications Act goal, but I think much more broadly speaking, having a robust Internet is something that is, I think we can all agree, beneficial for Canadians, period.
1152 COMMISSIONER KATZ: So, if that's a goal and you say I guess also in -- if I can find it here, and I can't...
1153 In a market-based environment, how does building out more capacity protect the interest of shareholders unless there's additional cost obviously and the accompanying additional price increases?
1154 Because my concern is you're saying, it's important to spend more money on capacity because that will create more innovation, that's fine, but short of increasing prices, if costs go up, the shareholders pay for it. Is that not the case?
1155 MR. GLICK: Marvin wants to take a shot at that. I think Rob has an intervention as well.
1156 MR. AMMORI: So, when we think of Moore's Law, which is, you know, every 18 months roughly the cost of electronics is cut in half or doubles in speed, and when you look at a market-based very competitive industry like the consumer electronics industry we see Moore's Law play out, you know, every year there's a new whiz-bang computer that's so much faster, has better graphics and every year it plays out and shareholders in these super competitive industries they get a return based on the innovations of their companies.
1157 And what we see in the telecom space is there's a lot less competition and the gains of that kind of innovation are passed on more slowly.
1158 So, you ask the question of, will prices have to increase based on investments in capacity? Not totally sure about that. There was a recent study by Pugh in the U.S. showing that the costs of delivering band width have gone down but the prices have gone up, at least in the U.S., have gone down around the world in terms of costs.
1159 And so what we see is we see that carriers are pricing perhaps not based on cost, it might be pricing based nearer to demand based on market power.
1160 And so one point is it's possible that prices will not go up if costs go up because they're already pricing above costs, it's not a cost-based pricing.
1161 And the second response I have is, I do think that increase in prices to ensure the radical and evolvable innovation of the Internet is worth the trade-off. I think in terms of global competitiveness and innovation and development, you know, slightly higher prices, if that is necessary, is worth the trade-off.
1162 And it's kind of like when you look at other markets, is it worth having more expensive pharmaceuticals if they're healthy, right?
1163 COMMISSIONER KATZ: Now -- and I hear you.
1164 MR. AMMORI: M'hmm.
1165 COMMISSIONER KATZ: And I expected that answer and so I'll throw this back at you.
1166 You represent consumer groups. You're saying the higher price is worth the trade-off. Do you have statistics to show that consumers are prepared to pay more money if that is the case in order to have a fully open as defined by yourselves network?
1167 MR. AMMORI: Well, I remember you asking this yesterday and I think the question is, you know, are you willing to sell the picture to buy a frame, right?
1168 What the consumers want is access to the open Internet, they want access to be able to attach any device, get any application and when you're asking, you know, what exactly do consumers want, consumers want the ability to go on Twitter. They didn't realize they wanted the ability to go on Twitter, go on...
1169 COMMISSIONER KATZ: But you don't have any studies to show that in a competitive sense they're prepared to pay five percent more, 10 percent more for value received --
1170 MR. AMMORI: We have -- millions of consumers have joined our coalition and my assumption is -- and I've never seen studies either way of would you pay five percent less to have a managed system by a network.
1171 But Rob has a point.
1172 MR. TOPOLSKI: In 1995 I bought a Pentium 150 megahertz computer for about $2,000. Today I can buy one that is 20 times as fast for about $200.
1173 This is the way that Moore's Law works. Consumers of electronics and goods in the technological space expect over time that technology will evolve faster than the price will increase and, in this case, the price actually drops.
1174 And we're focused on the network in this hearing, but it's very important to look at this in the entire computing communications ecostructure which includes more speed on the processor, more bytes stored on the hard drive, more pixels on the screen and more practical uses of our computers and networks, all without costing more money over time.
1175 COMMISSIONER KATZ: But we're here looking at whether there is an obligation by the network providers to spend more capital, I think what you're saying here, and you're saying it will likely lead to an increase in network capacity.
"Application-specific traffic management..." (As read)
1176 COMMISSIONER KATZ: And I'm quoting from paragraph 18 here.
"However, the natural effect of removing from ISPs the inappropriate crutch of application-specific traffic management will lead to an increase in network capacity." (As read)
1177 COMMISSIONER KATZ: So, technology works on a step function as you're probably aware. So, as a result, in order to do this, if there is an increased less efficient networks because of this -- I'm making the assumption there might be -- then there's a step function, someone's got to pay for it.
1178 The immediate result is either the shareholder assumes the risk or the higher cost in the short term will be passed on to consumers.
1179 I understand the long term, I understand Moore's Law, but in the short term there's a cost implication and I'm just wondering whether you have statistics to bear out the fact that consumers are prepared to accept the cost equation, if there is one.
1180 MR. GLICK: So, I mean, I don't accept the presupposition in the question that you're going to have a less efficient network merely because you prohibit application-specific traffic management because there is a whole universe of traffic management techniques, effective traffic management techniques that remain available to carriers, and this is the experience that Comcast has already gone through in the U.S., where they have subsequent to the FCC order started using application-neutral traffic management techniques and it hasn't all gone to, you know, heck in a hand basket.
1181 So, that's the first point.
1182 The second point is, we're not mandating adding capacity, that's not our submission. We think that adding capacity is better and it is historically what has driven innovation on the network, but we are not asking you to mandate it.
1183 We think that network operators should have their own incentives for adding capacity and we think ultimately adding capacity works really well.
1184 The last thing I will note is that you're asking for evidence of who would pay more for less. I mean, maybe the best evidence comes from the point that I made at the beginning of my oral remarks which is to remind us what the carrier ISP said in the New Media proceeding.
1185 COMMISSIONER KATZ: Okay. The reason I raised the issue of capacity is because you made a point of it both in your February 23rd and your opening comments this morning on page 5 under step 3, you say:
"This include first and foremost, adding network capacity." (As read)
1186 COMMISSIONER KATZ: So, I saw the word 'foremost'. To me that meant, the first thing you do is spend money to build up the capacity.
1187 MR. GLICK: It's key, but we're not asking you to be prescriptive on that point.
1188 COMMISSIONER KATZ: I guess the last question. Again, your opening filing on February 23rd, paragraph 34 you state in subsection (a):
"Today's protocols on the Internet currently already exhibit congestion control behaviours." (As read)
1189 COMMISSIONER KATZ: To your knowledge, are any of those application-specific?
1190 MR. GLICK: I'll let Rob field that one.
1191 MR. TOPOLSKI: So, most applications that transfer large amounts of data over the internet use the TCP protocol. So, that answer would be "no". That's not application specific that TCP protocol has the built-in congestion controls.
1192 They may -- the applications that their option may lay around additional congestion algorithms on top of that for their own purposes and convenience, but by and large TCP handles it and it handles it equally for all applications that use it.
1193 But there are some applications that use UDP or other protocols where the obligation to build in congestion control is then laid to the person who is using that protocol.
1194 It doesn't have any inherent or built-in capability to do -- UDP doesn't have any inherent or built-in capability to do congestion control, so the application designer has to take on that responsibility. And it is a responsibility.
1195 COMMISSIONER KATZ: Well, why does he have to take on that responsibility? Why can't it be totally agnostic?
1196 MR. TOPOLSKI: Well, we are talking about the application itself, so it is agnostic in that it only cares about itself or it's -- it doesn't -- it is the wear of the outside world and it is spelled out in the recommendations of the Internet Engineering Task Force that the responsibility to handle congestion control within a protocol or within an application using UDP belongs to the application designer.
1197 COMMISSIONER KATZ: Those are my questions, Mr. Chairman.
1198 THE CHAIRPERSON: Candice?
1199 COMMISSIONER MOLNAR: Thank you. I have a number of questions. I want to start with speaking of your virtuous circle that the open internet is a virtuous circle, the consumer, the application provider and the carrier.
1200 And we spoke about the potential of carriers incurring costs to either increase capacity or introduce traffic management softwares and investment in that. We spoke about metering consumers.
1201 What I don't understand is what are the obligations on the application provider to ensure that applications are provided or created, but innovation is done in an efficient manner. At the end, we want an open internet, we want to encourage innovation, but we also want an affordable internet.
1202 And so, how, under the tests you've provided and under what you've proposed to us, are there obligations or encouragements on innovators to ensure that these applications are created in an efficient manner?
1203 MR. GLICK: The short answer is that there is a highly competitive market in the application space. So, that's why it matters to application developers to go faster because consumers notice when applications are faster or slower.
1204 That's why, when you do a search on some of your favourite search like Bing or Yahoo, they often indicate the number of milliseconds that it took to do a search because they want to tell you precisely how good they are at returning quick results.
1205 COMMISSIONER MOLNAR: So, fast is one of the objectives, but how about efficient? How about efficient use of capacity?
1206 At the end, I do want to understand the PDP issue better and you've promised that you can do that for us. But that is something that has been brought forward to us potentially as what might be considered an inefficient use of bandwidth.
1207 MR. GLICK: Sure. And Robb is well-placed to answer that question.
1208 MR. TOPOLSKI: So, it's in absolutely no one's best interest, particularly that of the application developer, to congest the network. Nothing works well over a congested network.
1209 It frustrates both the purposes of the application itself and anything else that the consumer and members even then the consumer's neighbourhood might be trying to do on the network.
1210 So, there is a built-in incentive in that the network works best when congestion doesn't exist. It's a built-in incentive not to create congestion using their own application.
1211 And it's for this reason why they choose to use the TCP protocol to carry the freight, because that has built-in congestion control. It's the protocol that's used by almost all internet applications that carry any bandwidth whatsoever and competes fairly amongst competing applications, using the TCP protocol.
1212 COMMISSIONER MOLNAR: Sorry. I just want to understand what is their built-in incentive. If we're talking here that carriers need to increase capacity or consumers are metered for usage, what is the built-in incentive for applications providers to be efficient? I don't understand that.
1213 MR. TOPOLSKI: I can help with that.
1214 COMMISSIONER MOLNAR: And when I look here at your test, it would say it's just -- you know, it is discrimination to discriminate against an application. Could it be no unjust discrimination if it's an inefficient application?
1215 Because I don't understand where their incentive is in this, in this circle.
1216 MR. TOPOLSKI: In a case where someone's malicious software is interfering with the network, it moves from the area of bonified applications acting responsibly to the area of malicious programs interfering with the operation of the network and I would expect the network operator to take action in that case.
1217 I am not under the provisions of how to manage every day desired traffic over the internet, but how to attack an abusive use of the network.
1218 MR. GLICK: If I can just --
1219 THE CHAIRPERSON: That's not the subject of this hearing. I mean, we are not dealing with abuse. My colleague asked you for quite different: what are the incentives for a legitimate use?
1220 MR. TOPOLSKI: Well, the incentive for a legitimate use is that if an application was to congest the network, that application is not going to work well. It's not going to meet its goals, users will be dissatisfied and choose to use an application that works better, one that won't congest the network.
1221 It's not unlike the way that we as users of the highway choose our own times and methods of using the roadway so that we don't run into congestion ourselves.
1222 COMMISSIONER MOLNAR: Okay. I am having problems understanding.
1223 MR. TOPOLSKI: I can help you with that.
1224 COMMISSIONER MOLNAR: So, the incentive is if their application congest the network consumers won't use that application. But you have proposed that we put in place guidelines or a framework that would say you can't discriminate against an application, that you should -- you should put in place practices, traffic management practices that don't target specific applications, so their application will not be slowed more or less than anyone else's applications, as I understand it.
1225 So, I still don't understand where their incentive is. If they are going to be slowed down equally with every other application, their consumers will use it as they use everything else?
1226 MR. GLICK: If I can just jump in. I think Markham wanted to opine on this point as well. Maybe he can address this issue for you.
1227 MR. ERICKSON: I just want to introduce another way of looking at to try and answer the question you're asking because it's a very good question.
1228 So, to build upon what Robb said that the application is going to be designed to work in an open internet, obviously the application once users to be able to interact the application and have it work well. So, the question is, then, what about an application that might be causing congestion?
1229 It's not the application that's causing a congestion I think is the answer. It's because in some cases users, an application comes so popular with the users, it's the user demand that creates the congestion, right?
1230 So, it's not about how the incentive of the application provider because the application provider will develop an application that works in an open internet and if it becomes popular, there is the potential that there may be congestion in a certain network because so many users are demanding that application. And then, it becomes a question of how do we deal with user demand?
1231 I don't want the capacity issue to become sort of overdone here. That's where we have said simply historically network operators have dealt with the demand of their own customers by increasing the capacity to let their customers access those applications. The question was: will consumers pay more for a capacity?
1232 That's the story of the dial-up move to the high speed air-net move. Consumers move from paying less for dial-up speeds to more for high speed because they wanted to get those applications.
1233 So, we simply say that to deal with the user demand, there are some methods available to network operators and we wouldn't ask to be prescriptive in saying to network operators which one of those methods they should use, except they shouldn't deny users the choice of what applications they may choose to use.
1234 If certain applications are causing a lot of demand, the network operator can deal with that in other ways by limiting the users' bandwidth or charging more or providing disclosures, doing other things rather than arbitrarily picking out applications that just because they tend to be popular.
1235 COMMISSIONER MOLNAR: Okay.
1236 THE CHAIRPERSON: Let me just ask clarification on the basis of my colleague's basic question on the fact that you are allowing essentially application based.
1237 If I understood your test, you just say it's the absolute last resort. You should try everything else, you go through your analytical framework. If nothing else works and there is congestion, then indeed you can have application based restrictions.
1238 MR. ERICKSON: We would simply say "yes", that we are not saying that they're never acceptable, but they would have to go through, I think, a more rigorous test and an XNT test if it's application specific throttling because that presents potentially so much collateral damage.
1239 THE CHAIRPERSON: My implication then if there is a base for inefficient application and it does congest and there are no other ways to dealing with it, that designer of the application runs the risk of being accounted by an application base restriction.
1240 MR. ERICKSON: Under those facts, and there are tests. That's right.
1241 COMMISSIONER MOLNAR: Thank you, Mr. Chair. So, you would agree under your tests while it may be discriminatory, it may not be unjustly discriminatory when it's reviewed?
1242 MR. GLICK: Yes. I mean --
1243 COMMISSIONER MOLNAR: And that's the incentive for applications to remain efficient.
1244 MR. GLICK: And that's the statute and I would only add to that that there is a robust competitive market for applications and applications that are inefficient and don't work well on the network, won't be popular among users.
1245 COMMISSIONER MOLNAR: I just have one other question.
1246 Because you are very involved in this whole matter of traffic management, yesterday Juniper Networks was in front of us and spoke about approaches where the traffic management was in fact at the consumer level, that they were able to control it and they said that that isn't something that exists here in Canada today but technologies are available that would allow control to be at the consumer level versus the network level.
1247 Are you aware of this and do you believe that that's a solution?
1248 MR. GLICK: I'm going to ask Rob to answer are we aware of this, although I think the answer is yes.
1249 And whether it's a solution I couldn't say, but it's certainly something that we don't oppose. In fact, in some circumstances putting consumers in control of their own Internet connection is going to be the best way to so-called managed the network.
1250 But, Rob, I don't know if there is anything you want to add about the specific technologies out there.
1251 MR. TOPOLSKI: Well, it's a common thread at the Internet Engineering Task Force that the decision about instructions to send to network operators about the handling requirements for their packets rests with the end-user and the end-user's applications. That's the design of the Internet. That has been the design of the Internet since the late '80s. That the type of service bit in an IP header is set by the end-user and end-user's application in order to instruct how the packets ought to be treated through the network.
1252 Proposing a way to keep the end-user in control is completely consistent with that, although the specifics I would like to see at the IETF before they went forward. But it's consistent.
1253 I got the question is DiffServ an example?
1254 DiffServ is an example of -- it's a way for an end-user application or the end-user to instruct the network on how packets ought to be forwarded and it would be an example of the end-user being in control of those instructions to the network.
1255 It would be the network's option whether or not to follow those instructions, but it is not at the IETF. It is not the networks' option to do something else other than those instructions or to treat everything equally.
1256 COMMISSIONER MOLNAR: I get a sense -- and this is my final question just to understand that this is not something that's possible in the near term.
1257 MR. TOPOLSKI: I disagree.
1258 COMMISSIONER MOLNAR: I mean, you mentioned that the network is designed to work that way, but it doesn't work that way today. Today consumers don't control network congestion, that's why we're here.
1259 So how far away would we be from that?
1260 MR. TOPOLSKI: Well, it is a question of predictions and my prediction has been if the network operators would receive and respect the DiffServ markings from the Internet standards that within a year we would see many popular applications begin to use those markings and that the operators would begin to benefit those from those markings.
1261 COMMISSIONER MOLNAR: Thank you.
1262 THE CHAIRPERSON: Suzanne...?
1263 CONSEILLÈRE LAMARRE: Merci, Monsieur le Président.
1264 What about privacy? Isn't it a concern for U.S. consumers? Because you haven't touched on it at all during your presentation this morning and it's one of the issues that was identified as a discussion point here for this hearing.
1265 So my first question is: Is it a concern for U.S. consumers, because I can tell you it is a concern in Canada?
1266 And how can it be preserved in the open Internet that you are promoting here this morning?
1267 MR. GLICK: The short answer is yes, privacy is a concern, not just for U.S. consumers but for Canadian consumers and for Internet users globally.
1268 In the Canadian context, we note that the Privacy Commissioner is already seized with a complaint related to DPI and we think that she is well-positioned to look at that issue.
1269 COMMISSIONER LAMARRE: But our statute, you know, one of the objectives of the Telecom Act, section 7(i), calls for CRTC's Telecommunication Policies to contribute to the protection of privacy of Canadian citizens.
1270 MR. GLICK: So to the extent that you were to make a finding that deep packet inspection, for example, endangered the privacy of Canadian consumers, you might want to have that finding reflected in your ultimate order.
1271 COMMISSIONER LAMARRE: But that's not my question. The question I asked is how privacy can be preserved in the open Internet that you are promoting here this morning, because you are promoting an open Internet, you say let innovation go forward, but also innovation may have impacts on privacy.
1272 So my question to you this morning is: How can it be preserved in the open Internet that you are promoting?
1273 MR. GLICK: Well, I mean, in the Canadian context --
1274 COMMISSIONER LAMARRE: Yes...?
1275 MR. GLICK: -- privacy is preserved with a combination of private-sector incentives and public regulation and I think those are actually pretty effective. We have a good effective private sector privacy regime in the form of PIPEDA and I think that that helps protect Canadian consumers online today.
1276 COMMISSIONER LAMARRE: So there is nothing from the U.S. side that's being done that can be shared this morning?
1277 MR. GLICK: I'm sorry, there's no secrets about it. I mean the FTC and the FCC are both looking into all sorts of privacy related issues.
1278 I don't know if Markham or Marvin want to talk about that.
1279 MR. ERICKSON: Sure. Well, I think in this area in privacy in some ways Canada is further along than we are in the United States. The United States does not have a comprehensive privacy regime into space. So I think that's the short answer.
1280 But in the United States the FTC, Congress and the FCC are beginning to raise the questions that you're asking now and stakeholders are beginning to respond.
1281 The question in the context of managing networks and network congestion and intersection of privacy, I think any network management technique that involved the inspection of users communications and content or keeping profiles of users and what kind of applications they are using would raise privacy concerns and we wouldn't expect network operators to do that. But there may be an appropriate way for the Commission to set those guidelines out.
1282 COMMISSIONER LAMARRE: But if I may have a last attempt at insisting on this, you know, when you are talking about open Internet you are talking about applications. We will deal with the ISPs and how they are doing it, but right now I want to deal with what you are promoting this morning having an open Internet where you do not stop or slow down the pace of innovation, you let everybody try out the innovation.
1283 Don't you see there a potential for infringement on privacy and, if so, how do you think we could, either the Commission or the Internet itself, the system itself, prevent that.
1284 MR. GLICK: I mean, I think that any open Internet type innovation is still subject to law, so you can't have open Internet -- you can't have applications to put on the Internet that violate the law. So to the extent that we have good private sector privacy law in Canada, I think that is a fetter on the kinds of things that you might see deployed on the open Internet, and a good one, because the open Internet and innovation includes protection of privacy.
1285 COMMISSIONER LAMARRE: Okay. Well, I was hoping for a more technical answer, but I guess that's going to be it.
1286 MR. TOPOLSKI: If you are asking for a technical explanation I have been asked to provide input on technical questions.
1287 The issue is: Should applications be discriminated against based on their being an application or their conveyance of a certain type of content.
1288 From a privacy perspective you can only discover this by looking past the information that's given from the end-user in their applications for the routing of these packets --
1289 COMMISSIONER LAMARRE: Okay.
1290 MR. TOPOLSKI: -- into the content themselves.
1291 COMMISSIONER LAMARRE: I'm sorry to interrupt, but your answer is still directed to the network management.
1292 MR. TOPOLSKI: Yes. Yes, it is.
1293 COMMISSIONER LAMARRE: I was looking more at the application issues. So I think we are --
1294 MR. TOPOLSKI: Okay.
1295 COMMISSIONER LAMARRE: I think I'm going to let it go. Thank you.
1296 THE CHAIRPERSON: Okay.
1297 Just to wrap up on what my colleague just said on privacy, surely your step number three when you look at traffic management costs to whether they are least restrictive, there has to be an element of piracy consideration there.
1298 Does that measure invade privacy more than is necessary for the purpose, et cetera?
1299 So surely to that extent at least you are addressing privacy.
1300 MR. GLICK: Yes.
1301 THE CHAIRPERSON: And the other thing, you make this rather bold statement, Mr. Glick, on page 4 saying:
"We note, by the way, that it seems obvious that discriminating between applications is discrimination for the purposes of 27(2)."
1302 It is clearly discrimination; whether it is unjust or not surely depends on the facts. You can't just say that everything -- every application base is discrimination -- is automatically unjust.
1303 MR. GLICK: Yes.
1304 THE CHAIRPERSON: Okay.
1305 Lastly, on page 5 -- and again, you made this at least five times during your presentation, you say there is a whole host of application neutral based ITMPs that should be used rather than application-specific ones.
1306 You have until July 24th to make additional submissions, I would very much appreciate if you could list what those application neutral devices are. The only one I heard you mention this morning is TCMP, but if there is a whole host of them I would like to know what they are, because in our hearing so far the evidence has been largely that in order to do effective Internet traffic management you have to do some shaping and that involves discrimination between applications, particularly P2P obviously.
1307 But if there are other technical application neutral methods, I think you should outline to us.
1308 MR. GLICK: We will, Mr. Chair.
1309 Just to take an initial shot at that, I have asked Marvin to add something about what Comcast is doing.
1310 MR. AMMORI: With your permission.
1311 So when the FCC sanctioned Comcast for engaging in application-specific degradation against BitTorrent --
1312 THE CHAIRPERSON: Hang on. Comcast was discrimination between customers, so it was slightly different. It wasn't between applications.
1313 MR. AMMORI: It was between applications. Comcast was targeting certain applications including peer-to-peer --
1314 THE CHAIRPERSON: Yes.
1315 MR. AMMORI: -- and the FCC said you have to move to an application-neutral method and Comcast said -- Comcast within a few months, during the timeline from August to January, moved to an application-neutral system which at times of peak congestion, triggered based on, you know, 70 percent usage on the download or 80 percent on the upload, the user, who happens to be using 70 percent of their allocated advertised bandwidth, would then be given less priority than other users.
1316 So that was purely application neutral and Comcast seems to be very happy with that, with that method. So it just took a few months and they moved directly to that.
1317 THE CHAIRPERSON: Okay. So when you talk application-specific what are you talking about? You mean a specific application or specific type of application?
1318 MR. AMMORI: Either one. What Comcast did was it targeted specific applications --
1319 THE CHAIRPERSON: Right.
1320 MR. AMMORI: -- that were using the BitTorrent protocol or using the Nutella protocol.
1321 THE CHAIRPERSON: Right.
1322 MR. AMMORI: And when they went to application neutral they simply took the high bandwidth user -- say Mr. Katz, who's a gamer, high bandwidth use at the moment, right --
1323 MR. AMMORI: It would take Mr. Katz's connection --
1324 THE CHAIRPERSON: Yes...?
1325 MR. AMMORI: -- at times of peak congestion, if he happens to be using a lot of bandwidth, and make his second priority to your traffic.
1326 THE CHAIRPERSON: Right.
1327 MR. AMMORI: All of his traffic, not just the gaming application.
1328 THE CHAIRPERSON: Well, that's what I have, so applications.
1329 If every user gets slowed down, that's fine by you. Every user beyond a certain capacity, you just don't identify him by application, you identify him by usage?
1330 MR. AMMORI: Yes. That would be application-neutral --
1331 THE CHAIRPERSON: Yes.
1332 MR. AMMORI: -- and then you would have to look and see, you know, if it was overly burdensome.
1333 THE CHAIRPERSON: Right. Right. Of course, yes.
1334 Okay. Thank you very much for your excellent presentation. I appreciate that.
1335 As I say, if you can send me that listing, that would be appreciated.
1336 We will take a 10-minute break.
--- Suspension à 1013
--- Reprise à 1028
1337 LE PRÉSIDENT: Bon. Commençons, Madame la Secrétaire.
1338 LA SECRÉTAIRE: Merci, monsieur le Président.
1339 I would now invite Zip.ca to make it's presentation. Appearing for Zip.ca is Mr. Hall.
1340 You have 15 minutes for your presentation.
1341 MR. HALL: Thank you and I want to thank the Commission for the opportunity to speak before you.
1342 My name is Rob Hall, I am the CEO of a company called Momentous.ca which owns various Internet subsidiaries. They fall into two main groups. Certainly we all know largest domain name registries hosting and e-mail companies in Canada. Internic.ca and DomainsAtCost.ca would be two of our brands.
1343 We also own Zip.ca of which I am the Chair, Chairman of the Board.
1344 Zip.ca is in the traditional --
1345 THE CHAIRPERSON: Can you slow down, because the translators can't talk as fast as you can.
1346 MR. HALL: But I only have 15 minutes.
1347 MR. HALL: Sure.
1348 Zip is in the video distribution business by mail, so DVD rental by mail, similar to the Netflix model in the U.S. We ship about 40,000 DVDs a day currently. Of course that's the non-technical way of shipping a movie to a consumer at this point.
1349 I am also the former Vice Chair of the Canadian Association of Internet Providers, so I have been around this industry. I owned one of the first ISPs in Ottawa and I had the privilege of serving as the first Chair of the Canadian Internet Registration Authority, which of course governs the .ca domain.
1350 I am also involved heavily with ICANN, the Internet Corporation for Assignment of Names and Numbers, and sit on their Nominating Committee that picks their Board, as well as being the former Vice Chair of the Registrar constituency.
1351 So I am, by all intents and purposes, a geek, as my family describes me, and I'm happy to answer fairly technical questions, but I will try to keep it to business.
1352 I want to start by saying I was surprised to see in one of your definitions or assumptions about what this hearing was, a statement that the traffic generated by a telco or a provider -- a carrier if you will -- for their own services was exempted from this hearing.
1353 I think that's exactly the wrong assumption for you to make. It is exactly these services that they may be providing to their customers that Zip.ca is concerned about and worry that we may be discriminated against differently than they are. So if you are going to apply rules to how bandwidth is constrained or how services are delivered, you have to look at the individual telco or carrier services as well as ours, because they compete head-to-head. The same roles must apply to both.
1354 I would also say it's going to be tough to set rules globally.
1355 I was recently at the ICANN meetings in Sydney where they were looking at registrar/registry separation for example and how to set rules for that, should a registrar be able to own a registry, and it basically comes down to it's almost impossible to set rules globally. You need to make sure that you have something that allows you to look at the individual set of circumstances. Market power and non-competitive practices often are like porn, you know it when you see it, but it's hard to define it other than that.
1356 So I would encourage you, if you are sending rules, and I assume you are, that you make sure there is some way to identify infractions and settle them quickly. I will talk about the timing a little later.
1357 Zip is moving into the download area. So Zip.ca announced two weeks ago that we are one of the first to bring movies to Canada where you can download it onto a set-top box. We are integrating with many providers such as LG where it would be in your TV, Xbox. So we will be able to stream live -- or not life, but video content, first-run movies to your TV as one of the first in Canada to do so.
1358 This of course will use a fair amount of bandwidth, so you can understand our concern when we hear about filtering on an application level or discrimination against individual applications or indeed individual companies.
1359 The other thing I have heard bandied about over the last two days of testimony is deep packet inspection. This certainly worries us as well.
1360 We have a lot of competitive intelligence about what our users want. In fact, we often know what they want to view before they know it. So our recommendation engine is one of our strengths where we look at everything a user has watched and rated and liked in the past and we then recommend new movies to them and download them, if you will, to a set-top box before they even know they want to watch them.
1361 If deep packet inspection were allowed to look at the content of what we are downloading to our users, you can imagine how someone like The Movie Network or one of our competitors owned by the carriers would certainly want to know exactly what our viewers are wanting and use it to their competitive intelligence.
1362 The other concern we have is we hear a lot of times carriers argue that of course they won't abuse this or they won't use the rules to their advantage. We would say that's ridiculous. We heard VeriSign in the domain name world argue the same thing, that they wouldn't raise prices on .com by 7 percent just because they were allowed to.
1363 Of course they have to. If you allow them to, their shareholders will demand it. They are public companies, they have an obligation to their shareholders to work within whatever rules they can to gain advantage against us.
1364 The other concern we have is time to resolution. The Internet moves very quickly. You can imagine if we were setting up a download service, our consumers were downloading, our customers were downloading and watching video and all of a sudden the rules changed or a carrier changed the way we were able to deliver that; we can't wait months to resolve it.
1365 So we would encourage the CRTC that any rules you put into place, make sure you have some kind of a complaint mechanism and a quick resolution.
1366 I'm reminded of back to my CAIP days where, you know, things take years typically to be resolved. You can put a business, especially one like this, out of business very quickly by just simply not allowing us to deliver the service to our consumers.
1367 Innovation on the Internet is dramatic, quick and I would hate to see anything you do stifle that, and I would encourage you not to.
1368 I have heard a lot of talk about BitTorrent and how we must, you know, stop BitTorrent users from downloading illegal movies.
1369 It turns out, BitTorrent might be the ideal platform and protocol for us to deliver our movies legitimately with a digital rights management key to our end users. It's certainly the most network efficient to do it. So any program or protocol that allowed carriers to discriminate based on the type of application such as BitTorrent would ultimately end up harming consumers and our company.
1370 I will close with keep in mind that there is an "I" in the first of Internet. It stands for "International".
1371 I have heard a lot about walled gardens. Certainly that's not what we want to be creating here. We are close in proximity to the U.S. network-wise, we get our feeds mostly from the U.S. in terms of bandwidth, to put in a different regime, although it may seem very Canadian to do, could be a very dangerous move.
1372 The other thing to keep in mind is the "I" in Internet I believe also stands for me, the consumer. Ultimately it should be my choice what content I view, not the carrier's choice.
1373 Thank you. I will take your questions.
1374 THE CHAIRPERSON: Thank you.
1375 You said you were going to start now streaming down movies to your -- rather than sending people CDs.
1376 How does you activity differ from that of a VOD operator which we license under our Broadcast Regulations?
1377 MR. HALL: I gather in our model -- I'm not sure if you are talking businesswise or technically so I will answer both.
1378 Businesswise in our model, we are a subscription service, so you pay, for example, $25 a month and you can view all the movies you want, as opposed to a traditional pay $5.00, buy it now, rent it for an hour or two hours it takes you to watch it.
1379 We would also look at downloading to the set-top box prior to the user knowing that they even want to watch the movie, as I mentioned.
1380 Our recommendation engine is very good at saying: Hey, you liked this TV show, or this movie, we think you will like that one. I will download it overnight and have it waiting for you, ready to go.
1381 On a business level, I think we are a little different than your typical video-on-demand --
1382 THE CHAIRPERSON: You would offer both subscription video and one-time video-on-demand.
1383 MR. HALL: Yes, we would offer both services.
1384 THE CHAIRPERSON: So you are basically doing exactly the same thing that the licensed VODs over cable do, except you are using the internet.
1385 MR. HALL: I would say that we offer a very similar service, yes.
1386 THE CHAIRPERSON: And it comes down on my computer, presumably, and then I have to get it from my computer to the TV.
1387 MR. HALL: No, sir. The deal we announced with CinemaNow and Sonic in the U.S. -- we are bringing the rights to Canada. So we are in the process of negotiating different rights to Canada. The challenge has been, of course, that the studios in Canada hold different rights than in the U.S.
1388 THE CHAIRPERSON: Right.
1389 MR. HALL: With that deal, we will be on set-top boxes, built into devices such as the Xbox, LG devices, both DVD players and TVs.
1390 So it will already be on your TV.
1391 That is one of the biggest consumer problems, no one wants to watch a movie on their computer, they like to see it on their big plasma screen.
1392 THE CHAIRPERSON: But I would need a separate set-top box, like you said, the Xbox or something like that.
1393 I mean, my set-top box from Rogers wouldn't allow me to receive it, unless you did a deal with Rogers.
1394 MR. HALL: That's correct. Rogers uses a box, typically, by Scientific Atlanta, which is a proprietary box. No, we would not be allowed by Rogers to stream to that box, although that box does talk TCP/IP and is plugged into the internet.
1395 But it is a proprietary issue, I believe.
1396 THE CHAIRPERSON: You made some comments about the assumptions. You said that our assumptions are wrong because we are only talking about the free internet, we are not talking about the carrier internet, customer-specific.
1397 I am not quite sure that I understand. To the extent that carriers offer an open service -- let's say that Bell offers Sympatico or something like that, or gives you ADSL access, et cetera -- they would be included, so where is our error?
1398 You said that we made a mistake by not including ILEC carriers' specific traffic. Those were the words you used.
1399 MR. HALL: Let me be very clear on that. I believe that your definition is that the term "internet traffic" refers to data carried on the public internet. It exempts traffic that is carried on a service provider's network, but not on the public internet, such as a licensed IPTV service or a managed Voice over IP service, and that is not considered public internet traffic.
1400 It is exactly those types of services that will be competing with ours. If you are going to set rules for the internet traffic, the public internet traffic, that we are affected by, the carriers should be affected by them, as well, for their own services.
1401 THE CHAIRPERSON: I'm sorry, I don't follow you.
1402 Licensed IPTV is basically cable over telephone wires; right?
1403 It's like an ILEC. Instead of Rogers, I subscribe to the IPTV of Bell --
1404 MR. HALL: Sure.
1405 THE CHAIRPERSON: -- and the signal comes from a telephone wire rather than a cable wire.
1406 I can buy a basic package, I can have options, blah, blah, blah, blah.
1407 How is that in the open internet?
1408 That is a specific service that that carrier sells to me, the same way as Rogers sells it to Len, who happens to be a Rogers customer.
1409 How is that in competition with you, who are operating on the internet, and you are not regulated, you can do what you want? You deliver what you want, in whatever format you do it.
1410 MR. HALL: If it is carried over the TCP/IP lines that my data is, and they prefer their traffic over mine, it's a problem, sir.
1411 If they are offering a service to end consumers over their network that is the same network that I am trying to offer a similar service to them on, and they are not governed by the same rules, and they are able to discriminate against my traffic and not theirs, that is exactly the root of the problem.
1412 THE CHAIRPERSON: What exactly do you mean by "discriminate" in this context?
1413 MR. HALL: If their internet traffic is exempt from any rules you set on network neutrality, your application level filtering, then basically they are allowed to say that their application goes through with priority and mine doesn't, and that certainly would be an issue to us.
1414 THE CHAIRPERSON: Therefore, what is your suggestion?
1415 I am not sure that I understand that that is what will happen, but assume that happens; therefore, you say that whatever rules we devise --
1416 Just before you we heard from the OIC, who suggested that we adopt a certain protocol and that we examine all the measures they are proposing. We should, in so doing, include the traffic that they produce over their licensed IPTV?
1417 MR. HALL: Correct.
1418 To the consumer it is the same end service. Allowing them to filter mine differently than theirs will produce an advantage to them competitively.
1419 THE CHAIRPERSON: Except that IPTV is -- in your case, it's the same. The internet has thousands of other uses which have absolutely nothing to do with watching movies.
1420 The approach was that we want to make sure there is an open internet, as the people before you said, that anybody can go on and innovate, et cetera. When there is congestion, apply rules that are neutral, so you don't discriminate.
1421 MR. HALL: Correct.
1422 THE CHAIRPERSON: Now, you say, in the specific case of, essentially, an internet VOD service like yours, you are actually competing with IPTV.
1423 MR. HALL: I am saying that you cannot allow them to discriminate and exempt their services that compete with those services they are allowed to discriminate in ways against.
1424 If you are going to set rules that allow them to manage network traffic -- and hopefully you set rules that don't allow them to discriminate at the application level -- to completely exempt their applications, as if they are somehow different, when they aren't, technically, is a problem.
1425 THE CHAIRPERSON: The OIC, before you, said: Don't do application-based, do consumption-based traffic management.
1426 So if anybody uses over 80 percent capacity, or whatever it happens to be, you can start throttling or slowing down.
1427 You say that, if you do that, you also slow down the IPTV.
1428 MR. HALL: Yes, please.
1429 There is an issue with that. The danger there is -- I will give you a classic example. Let's say a video to download is 9 gigabytes, for example, for a DVD. If the carriers were allowed to say, "We will start throttling when you hit 8 gigabytes, but we don't have to throttle our own service, and we can deliver all 9 to you at high speed," that would be a problem for us.
1430 So to say that you can throttle or you can discriminate just based on volume can also be dangerous, depending on where they set those caps.
1431 If they set them just below where we are allowed to deliver the service that our consumers want, but their own services can deliver those services free and clear of any rules you set, that would certainly be an issue for us.
1432 So while we are typically in favour of volume-type caps, because that makes the most sense network-wise, we have to be careful that they are not set in such a way as to discriminate against the applications -- the large applications, if you will, such as ours.
1433 Because video will be a larger application than, say, e-mail, for example.
1434 THE CHAIRPERSON: You are really worried about measures that, on their face, would appear neutral, but, effectively, they are not and they would have a different effect.
1435 MR. HALL: Correct. Absolutely.
1436 THE CHAIRPERSON: Suzanne, do you have some questions?
1437 COMMISSIONER LAMARRE: Yes, thank you, Mr. Chair.
1438 I just want to understand properly your business. I understand that it is subscription-based, and, also, the way it is operating right now is through the mail, by rentals, you get the DVDs. Do you also engage in selling those DVDs?
1439 MR. HALL: I think the simple answer is "Yes". We do that in a couple of different ways. It is certainly not our primary business, but when we buy, for example, 2,000 copies of "Troy", which is a wonderful movie, and we have 1,999 of them sitting in our warehouse after two weeks, we do sell the used copies, and I believe that we do make some new releases available for sale to our clients at a discounted rate.
1440 COMMISSIONER LAMARRE: But that is not the main purpose of your business.
1441 MR. HALL: It would be less than a 10th of 1 percent of our revenue or business.
1442 COMMISSIONER LAMARRE: Okay. Well, then, let's not even mention it.
1443 You did mention in replying to our Chairman that one movie, for example, could take 9 gigabytes to download. Is that an average size of bandwidth that you would need to transmit to consumers the movies, or whatever programs they want to watch?
1444 MR. HALL: No, that would be a typical, standard DVD. Certainly with compressions and new codecs coming out to compress movies, we can get it down below 1 gigabyte even.
1445 So it depends on the resolution.
1446 Of course, Blu-ray kind of throws that all up in the air. If you are going to try to download a Blu-ray movie, then you are into 40 or 50 gigabytes before you start compressing it.
1447 I would be hesitant to say an exact number, because it changes so rapidly. Every day we are coming out with new compression technologies that allow us to deliver these services differently and more efficiently, and it is certainly in our interest to do so.
1448 But that was an example of -- you have to be careful that they aren't setting the barrier just below where you can deliver a service.
1449 COMMISSIONER LAMARRE: I understand.
1450 Currently you have not yet started operating in that mode of downloading. You are still operating in the mode of using the postal service to get the DVDs out to your clients. Correct?
1451 MR. HALL: I would say that it is currently not in production. We do download trailers at this point, where you can watch a movie trailer before you decide to order it.
1452 Over the last two or three years we have been streaming, for example, NHL games and other types of video under licence from the NHL, as what I will call a trial, for lack of a better word.
1453 For instance, we had a deal with the National Hockey League that allowed us, a day after a hockey game was played, to archive it forever and display it on our website.
1454 So we have been doing that, but I would classify those more as some trials that we were doing with different content holders and rights holders, to get our systems up and running.
1455 COMMISSIONER LAMARRE: I did go to your website --
1456 MR. HALL: Oh, thank you.
1457 COMMISSIONER LAMARRE: -- and I noticed that, in marketing your services, you say that there are no postal fees, you know, "We will get you the DVD."
1458 Obviously, being a businessman, you want to make a profit, so you have factored in the delivery cost of getting the DVD to the customer and back to your shop. You have factored that into your subscription price.
1459 MR. HALL: Yes.
1460 COMMISSIONER LAMARRE: Going from the postal service to downloading on the internet, what you are actually doing is -- basically, you are changing the delivery mode, and, hopefully, you will end up with a delivery system that will be faster and more convenient for your clients.
1461 How do you expect that the cost factor of the delivery will translate into the new mode? Do you expect that it will end up being the same expense for your enterprise, or do you expect that it will go down or up?
1462 MR. HALL: Right now it is higher than delivering by mail. It costs us more to download a high-res video -- in our bandwidth cost; not the consumer bandwidth cost, but our end of the bandwidth cost.
1463 Right now it costs more for us to download the video than to ship it by mail.
1464 Will that always be true? I certainly hope not. Bandwidth costs tend to drop very rapidly over time.
1465 The problem with the theory of "Hey, this will all be great in a year," is that, of course, as bandwidth costs drop, consumers demand higher and higher resolution, such as high-definition and Blu-ray.
1466 So although it would be great for us to say that a year from now we hope that bandwidth costs would allow us to do it cheaper than mail, at this point that is not the case, it would cost us more to deliver a movie with bandwidth than it would to deliver it by mail, which is a little perverse in our sense, if you will, but that is the world we live in.
1467 In the U.S. this wouldn't be true. We can buy bandwidth much cheaper in the U.S. than we can in Canada. However, we have always decided to keep our video servers in Canada.
1468 COMMISSIONER LAMARRE: In paragraph 16 of your December submission you say in the first sentence, "Zip.ca's primary concern is the potential negative effects that traffic management or shaping may have on our ability to provide paying subscribers with consistent high-quality service," and we are talking about the high-quality service of video.
1469 That being said, do you know of any internet traffic management practices that currently exist, either here or in the U.S., that affect video quality, either in a good manner or in a harmful manner?
1470 MR. HALL: I certainly have anecdotal evidence. We don't have any solid evidence of that happening.
1471 I can tell you that we have been trying to peer with some of the carriers at Torex, one of the Toronto internet exchange points, and have been unsuccessful in doing so. Whether they feel it is because we will be dumping a lot of bandwidth down to the subscribers, or whether they feel that they would rather have us pay for a connection to them, I can't speak to their motivation.
1472 So, yes, we have had trouble connecting to the carriers because we are Zip.ca and they know that we deliver video.
1473 Interestingly enough, what helps us in some cases connecting to carriers is the other side of our business, our domain registrar and hosting and DNS services, which are required services that they want.
1474 However, when they hear we are Zip, the conversation typically changes to one of negativity and, "No, we don't want you to peer and connect to us, we would like you to buy bandwidth from us."
1475 COMMISSIONER LAMARRE: Okay. Those are my questions.
1476 THE CHAIRPERSON: Tim...?
1477 COMMISSIONER DENTON: Just for the sake of my curiosity, how much does it cost you to mail out your movies?
1478 MR. HALL: We pay postage both ways on our movies. Zip is a flat-rate service. We charge $25 a month, typically, for four movies out at a time. You make a list from our 72,000 movies of which ones you want to watch, we pick your top four and mail them to you. When you are done one, you throw it in the mail back to us, and we pay postage both ways.
1479 That is a 10-second summary of our service.
1480 I think it's about 52 cents currently -- 51 or 52 cents. So call it $1.04 both ways. At 40,000 movies a day, you can figure out what our postage rates are. We are one of Canada Post's top customers, and we are, interestingly enough, one of the few customers that Canada Post has that is growing, and trying to find more ways to send things by mail.
1481 MR. HALL: Although, one day I hope that we actually find ways to send things by internet rather than mail.
1482 We have various handling fees on top of that, obviously. It costs about 23 cents internally to unpack something and repack it and send it to another customer from our four distribution centres across Canada.
1483 We operate out of Ottawa, Toronto, Calgary and Vancouver currently.
1484 COMMISSIONER DENTON: And you are asserting, just to hear you say it again, that all of that is still cheaper than sending it out by broadband -- electronically.
1485 MR. HALL: At today's bandwidth costs in Canada, yes.
1486 COMMISSIONER DENTON: Thank you.
1487 THE CHAIRPERSON: Those are all of our questions for you. Thank you very much for your appearance.
1488 MR. HALL: Thank you.
1489 THE CHAIRPERSON: Madam Secretary, let's deal with the next presentation.
1490 THE SECRETARY: I would now invite the Coalition of Internet Service Providers to come to the presentation table.
1491 THE SECRETARY: Appearing for the Coalition of Internet Service Providers is François Ménard.
1492 Please introduce yourself for the record, and begin your 15-minute presentation.
1493 MR. MÉNARD: Good morning. I will do my presentation in English, since I have written the speech in English.
1494 My name is François Ménard, and I work for a company called Xittel. We are one of the members of the Coalition of Internet Service Providers, which, as you will know, has participated in many CRTC proceedings over the last few years, and we are very concerned with the issue of internet traffic management.
1495 There is not very much of the internet in there, it is actually IP traffic management, because most of the issues that are being analyzed are not on the "inter" portion of networks, but actually within a given carrier's access network, mainly.
1496 When this thing is looked at in perspective, the focus of this analysis is the extent to which there are section 36 violations in the way that traffic is being managed, typically on access networks, and since today peering between ISPs is forborne and has not been the subject of any Commission scrutiny for many years already, calling this internet traffic management, I think, is a misnomer and it is worth mentioning that it is really IP traffic management.
1497 Our intent was to boil down our position into three key points, three very simple points. The first one is encryption. Encryption makes deep-packet inspection pretty much ineffective, and encryption is pervasive.
1498 The second point is, it is all about congestion signalling, telling your network that it is congested to applications, or the lack of that, which is causing people to deploy deep-packet inspection and throttling.
1499 The third point is, in the end, this will all be solved with wire-speed aggregation, and if aggregation is done without any overbooking, like selling more seats on the plane than there are seats on the plane, then there will be no need to engage in traffic management as an "on behalf of" function, where the upstream service provider, the incumbent owning the pipes, feels compelled to do it on behalf of, say, the wholesale ISP.
1500 Those are the three points that I am going to dive into in my presentation.
1501 Today on the internet the expectation is that the security of telecommunications is to be achieved through the systematic use of cryptography to ensure proper authentication and integrity of internet applications, as well as, simply, privacy.
1502 And I heard on the way in that there were some interests on the part of the Commission to look at how privacy is maintained in internet applications. Flat out, if you want privacy on the internet, you encrypt. You encrypt, you make it impossible to engage in deep-packet inspection. If you make it impossible to engage in deep-packet inspection, there will be congestion.
1503 It is kind of a circle that you can't really get out of until you start looking at internet traffic management that works effectively in the presence of encryption.
1504 Most internet users by now have been the victims of spamming, fishing, spoofing and viruses, and have since stopped making use of applications which have not been upgraded to support strong cryptography.
1505 The restriction on the use of the now gigabyte lanes that we have on the internet, with fibre optics and so forth -- the restriction on the use of these lanes to, simply, applications which are not secure, so that they can get on these lanes -- for if they are secure, they will be shaped -- is not in the public interest. It took us many years to get to the point where we have a secure internet, and this is not something we want to lose. It costs a lot to society to take care of computers which are infected. It causes denial of service effects, puts down banking networks, and paralyses the internet.
1506 So we want a secure internet, and we want traffic management that works in the presence of encryption.
1507 Internet service providers are generally unable to isolate the underlying internet application which is the cause of congestion when the network traffic is encrypted.
1508 If ISPs are able to isolate such applications, then the encryption is no good and will be quickly replaced with a stronger one.
1509 Like anti-virus software on computers, traffic shaping hardware is sold with a mandatory subscription for new definition files, and failing to update your traffic management hardware for these new definition files will mean that your network will let through applications which will lead to network congestion.
1510 We simply submit that until such time as incoming carriers are required to demonstrate to the Commission -- and they have not yet -- that their deployment of deep-packet inspection and throttling for peer-to-peer traffic has been effective at eliminating network congestion, the Commission should not presume, for the evidence submitted by CISP in this filing, that as soon as 1.5 percent of users engage in peer-to-peer-type applications, it is enough to cause congestion, and if they are encrypted they will not be detected. Therefore, there will be, at least, 1.5 percent of users engaging in peer-to-peer traffic which will cause congestion.
1511 So the Commission should not presume that deep-packet and throttling has been effective at eliminating the presence of congestion. There is no evidence of that.
1512 The second point is simply that it is all about congestion signalling. If there were requirements for carriers to let applications know that their network is congested, applications could actually have the means to listen to that information and comply -- back off, slow down.
1513 But I don't know, including us as a carrier -- we don't necessarily want to tell in the open where a network is congested. Obviously this is information that we don't want out in the open, otherwise customers would go to other people.
1514 So we are kind of caught in this circle where we don't want to tell that our network is congesting, we don't want to tell where, and we want to find means of managing traffic such that we don't have situations where the network will congest, especially not for low value, high bandwidth applications such as use of peer-to-peer, which you would be more than happy --
1515 I use peer-to-peer all the time but I don't care if it takes two weeks for something to download sometimes and it doesn't have to be downloading at full speed within three minutes and break the network while this is taking place. And I certainly do not, as a user of peer-to-peer, expect even if I download 30-40-50 gigabytes of stuff in a month to be penalized or be told that I am endangering the network integrity when I am not really, I am just downloading lots of data really slowly.
1516 So this notion that a quota is measured in gigabytes per month instead of gigabytes per day or gigabytes per hour is kind of arbitrary and very statistical historically in the duopolistic environment.
1517 This issue was brought in detail in the past before the Commission, back in the cable modem tariffing days where the cable carriers introduced per gig quotas on TPIA tariffs which had no cost studies and still to this date have no cost studies backing the per gig rate.
1518 So the Commission took kind of an easy way out on this and kind of said that it would not be discriminatory if the per gigabyte overconsumption rates matched the retail rates that cable carriers charged their end users.
1519 And really there is a big difference, is that us as ISPs were forced to pay these rates whereas the cable carriers, when a customer complains and wants to switch and says, you know, I am not going to pay that or else you are going to lose my business, then they can simply waive these charges and they will not waive these charges to competitors.
1520 So there is sort of a misconception that deep packet inspection is evil and Internet traffic management is a good thing. Actually it is the other way around.
1521 I listened to people from -- a vendor yesterday explained that deep packet inspection actually is a good network integrity protection tool that pretty much acts like a firewall to prevent uses of networks beyond ways that the network use was intended to do, like support denial of service attacks and so forth, to block that.
1522 So deep packet inspection is inherently a firewall and a passive kind of thing. What is really active and which is taking action is the Internet traffic management or the IP traffic management portion. That is what can potentially violate section 36 of the Act.
1523 Our position is that network congestion can be mitigated by the use of Internet traffic management technology that is capable of communicating with Internet applications and instructing them to slow down, while at the same time keep a whip in hand and being able to enforce acceptable use policies and being able essentially to increase the pace at which packets will be dropped for end users who are not able to keep a tight control over their applications which are going rogue.
1524 So CISP has provided the Commission with detailed evidence of a technology developed by a company founded by Dr. Lawrence Roberts, one of the four founding fathers of the Internet, called Flow Management.
1525 The technology focuses on signalling for the presence of congestion and selectively dropping IP packets after precisely measuring flows and dropping packets no more than once per round-trip time. It is kind of a technical kind of thing which says that within a transaction there will be no more than one dropped packet and this is sufficient for all operating systems out there to listen and comply with.
1526 In other words, upon one packet being dropped, Windows, Linux, OS X, has a TCP/IP stack which will monitor a packet being dropped and slow down. You don't have to drop at will or engage in wild traffic management to induce a simple message: My network is congesting, slow down.
1527 And so because flows are being measured only from the Internet Protocol header, Flow Management remains entirely effective in the presence of encryption. I spoke about that earlier.
1528 So when incumbent carriers provide a transport service transparent to the Internet Protocol, the actual messages of congestion signalling must be relayed by the ISP towards the users of the ISP.
1529 There was lots of evidence in this proceeding that demonstrates that, for instance, the ILEC DSL wholesale network architecture is not an IP network, it is a PPPoE transport network, and the traffic-shaping hardware that the incumbent telephone companies use has to open the box, look at the PPPoE header, open the box again, look at the IP header, open the box again, look inside the IP packet to make a decision.
1530 It is extremely processing-intensive and ultimately it is a responsibility that does not behold upon anybody who is not the ISP of the end user. And in this case the incumbent telephone company is not the ISP of the end user. It is simply transporting traffic for another carrier and it is that carrier, i.e., members of this Coalition, which have the fiduciary right to signal for the presence of congestion.
1531 But we are not being told by the incumbent phone company or cable company where the network is congesting. So we can't tell applications to really back off.
1532 So it is kind of an in between situation where we would like to be part of the solution and we are not being offered the chance of being part of the solution.
1533 So rather than to cost your wholesale services in such a fashion as to allow ISPs to self-supply your own wire speed aggregation, certain incumbents have found it proper to engage in traffic shaping on portions of their backbones which are not even congested, while at the same time refusing to communicate to ISPs in real-time the state of congestion on the rest of their networks.
1534 So in doing so, they have failed to offer ISPs the opportunity of being part of the solution and therefore can only put the blame on themselves for having failed to upgrade their networks to be capable of delivering wire speed aggregated wholesale services.
1535 So CISP fundamentally believes that the process set by the Commission in Telecom Notice of Consultation 2009-261 will make it possible to outline a proper framework for enabling ISPs to self-supply their own transport to incumbent carriers' central offices and cable head ends, while relying on minimalistic regional wire speed packet switching and aggregation to be performed by the incumbent on behalf of the ISP, rather than to rely on aggregation embedded in the monthly rates, which does not supply the necessary bandwidth to properly offer services and leading to a situation where the only way out for these incumbent carriers is to start throttling.
1536 They are just not simply delivering aggregation bandwidth in the quantity required and this was demonstrated with a nice chart in our submission, where we have shown the growth in access network speed and we have shown the growth in backbone speed and we have shown the growth in aggregation network speed, and that curve was not nearly close to being on the same slope. There has been heavy under-investing in aggregation and that is what has sort of led to the situation we have today.
1537 I want to be very clear about the concept of wire speed aggregation, what it means. So I have thrown in there the definition for wire speed from Wikipedia:
"Wire speed... refers to the hypothetical maximum data transmission rate of a cable or other transmission medium (such as wireless). The wire speed is dependent on the physical and electrical properties of the cable (or wireless medium), combined with the lowest level of the connection protocols.
When used as an adjective, wire speed describes any hardware device or function that processes data without reducing the overall transmission rate." (As read)
1538 That means that on any router switch, DSLAM, CMTS, the sum of all downlinks towards end users must be totalled up and equate to at least the size of the pipe going out of the box so that the box itself never is a choke point.
1539 And if you aggregate multiples of such devices, say, in a head end or a central office, then the switch which gets traffic from many DSLAMs or many CMTS -- cable modem termination systems -- must have an egress outport, say, 10-gigabit ethernet that is at least as fast as all downlinks towards all the CMTS and DSLAMs.
1540 If you have that and if you deliver a big fat pipe which is not congested to the ISPs, the ISPs will be fire-hosed in the face with a huge quantity of bandwidth. They will be the entities which will have the onus to engage in proper traffic management.
1541 But since the underlying network architecture will be cost such that there is no overbooking in the cost structure of the underlying network architecture, then we will not have a situation where the incumbent carriers feel compelled to engage in traffic management on behalf of the ISPs.
1542 It is not as bad as it looks. There are simple ways of costing this out and this will be all done in 2009-261.
1543 So we would consider that regionally aggregated DSL and regionally aggregated TPIA offered and aggregated in 10-gigabit speeds would represent an acceptable form of what we would consider to be wire speed aggregation and packet switching.
1544 The best example we can use to compare this against is how when you are a CLEC and you interconnect with a telephone company, you don't get busy tones. It's not like the network is designed in such a fashion to be overbooked when you interconnect. Yet, as a CLEC, we went from 2,800 telephone exchanges in Canada to 280 local interconnection regions which are regionally aggregated points of interconnection.
1545 The old model where the ISPs were provided wholesale services on a silver platter has failed and 10 years later we are still debating the same topics. Until we change this architecture, then we will not solve the problem of Internet traffic management properly.
1546 So on this note, we would simply recommend that the Commission encourage the industry to implement network congestion signalling mechanisms for Internet applications, which are simply and only actually triggered by the presence of congestion, for without congestion there is really no need to discriminate against applications or discriminate against peer-to-peer applications.
1547 And even if there is a requirement to do that, which there is until we have more bandwidth out there, there are ways to make this effective by looking at technologies such as Flow Management, which will work in the presence of encryption.
1548 So the last thing that is really required is to make sure that the incumbent carriers are not allowed to throttle on behalf of the ISPs on the wholesale network architecture.
1549 That is my conclusion.
1550 THE CHAIRPERSON: Thank you very much for your presentation. I must say I read your submission last night in great detail. For a non-engineer, non-expert, it is very heavy reading. I mean it is replete with jargon and technical expressions which I am sure are very clear to you. For me, as a non-technician, it is very difficult to absorb.
1551 But I understood basically you are quite critical of us. You are saying that we are using the wrong data. We should not be using monthly data but much more accurate data. You suggest that focusing on DPI, which you say is not very useful because DPI is an outmoded tool if you can get around with encryption. And you also say that this hearing should have been combined with a next-generation hearing because the issues of the two are really interconnected and you can't look at them separately.
1552 Did I get that right?
1553 MR. MÉNARD: M'hmm.
1554 THE CHAIRPERSON: That is basically the tenure? Okay.
1555 Then you start off by quoting Dr. Roberts and suggest that rather than throttling and doing things that way, you should adopt what Dr. Roberts has outlined and you quote him at great length in paragraph 22 of your submission.
1556 Do me a favour here, on a keep it simple stupid basis, explain to me what Dr. Roberts suggests should be done.
1557 MR. MÉNARD: Sure. Dr. Roberts has been thinking about this stuff for longer than everybody in this room, you know, since 1964, and he has had lots of time to work out the details.
1558 What he has found is that -- and if you will accept that this will take about 30 seconds to explain -- is that if you are at home behind your DSL modem or cable modem and you are downloading so fast that at some point in time you congest the network, then there is basically a natural barrier to download faster, it is the speed of your modem.
1559 But it is in the other direction where the problem really is because today we have very asymmetrical networks where the upload is one-tenth as fast as the download and if you congest the upstream channel, then you can't click to download any more stuff.
1560 And because the characteristic of peer-to-peer applications is to upload data to other users, when these applications take control over your upload, you basically deny access to the entire capacity on the download side.
1561 So essentially the speed of the network in both directions ends up being only as fast as your capacity to upload data because you are denied access to the full download capability.
1562 THE CHAIRPERSON: The upload becomes the lowest common denominator?
1563 MR. MÉNARD: Correct.
1564 THE CHAIRPERSON: Yes.
1565 MR. MÉNARD: And that takes place as soon as 1.5 percent of end users are engaged in peer-to-peer applications. That is the evidence he has submitted.
1566 So we have a big problem in that, first off, we have networks which are highly asymmetrical right now with DSL and cable modem, but the same behaviour takes place in symmetrical networks. It's still valid. It just affects the percentage of end users.
1567 So what we really need to do is stop relying on mechanisms that need to look inside the packet in order to discriminate against peer-to-peer applications.
1568 THE CHAIRPERSON: And instead do what?
1569 MR. MÉNARD: And instead do what? What you basically do is you look at the distribution of traffic. Say my computer is speaking to 50 computers, uploading data at roughly 50 kilobits per second, and therefore this looks like and smells like a peer-to-peer use and clamp on that.
1570 In the past, routers were designed to only look at a packet one packet at a time and make a decision, okay, here is a packet, send it here, oh, here is another packet, send it there, but never remembering what was done in the prior transaction.
1571 THE CHAIRPERSON: M'hmm.
1572 MR. MÉNARD: Today, with faster routers, you can actually measure the pace at which packages get inside the postal office. Instead of relying upon the postal office itself to put the boxes in the higher priority truck and then be overloaded by packages that get into the postal office, you pace the speed at which the packages get inside the postal office so the postal office can do its job properly.
1573 That requires new technology and this is why the evidence submitted by Dr. Roberts is novel and relevant here because it's the only middle approach to traffic management, which ultimately delivers on a notion of signalling for the presence of congestion. If applications are told to slow down, they will, and if they don't, then you take the whip out and do something but you have to try to tell applications to slow down first.
1574 THE CHAIRPERSON: Okay. But is that technology available today?
1575 MR. MÉNARD: Absolutely.
1576 THE CHAIRPERSON: Why do the ILECs not use it or the ISPs? Why are they throttling? Why are they focusing on P2P? Why don't they do exactly what you are suggesting, look at the application and see what type of it and look at --
1577 MR. MÉNARD: Because it works at the IP Protocol level and currently the ILECs are not IP transport networks, they are PPPoE transport networks.
1578 THE CHAIRPERSON: Well, you are losing me when you use those acronyms.
1579 MR. MÉNARD: It is very important and I am emphasizing that and there are staff of the Commission which can really work these details out.
1580 THE CHAIRPERSON: I am sure but I have to make the decision, so I have to understand what you are talking about.
1581 MR. MÉNARD: Absolutely!
1582 THE CHAIRPERSON: So explain to me what you are talking about.
1583 MR. MÉNARD: Okay. So, in other words, the incumbents use a protocol for transporting traffic that could transport any other thing. It does not -- the ILEC DSL wholesale network can transport IPv4, IPv6, different protocols, but what we need here is a technology that works on the IP protocol and the technology that Flow Management is based upon, dropping packets such as to induce notifications of the presence of congestion.
1584 The incumbent phone companies cannot rely on hardware that isn't capable of looking at just an IP packet. They have to open the PPPoE box first. So they need this technology that looks and first opens the PPPoE header and deals with the content inside. Actually it is even more complicated. It is PPP inside L2TP and so forth. But the point being here is that the technology that the ILECs use places a greater demand on processing for looking inside the packet than that which a company like Anagran delivers today.
1585 However, this technology is not meant to be deployed in the ILEC network architecture, it is meant to be deployed on the ISP side so that the ISPs can tell the end users' applications to slow down.
1586 THE CHAIRPERSON: So if I am Bell Canada and I agree with you a hundred percent, that is the solution, what does that mean, they have to buy a new technology for their entire network, that the one they that they have right now does not do the trick, if I understand you correctly? And partially, which is your second point, is if they work together with the ISPs, part of this work can be done by the ISPs?
1587 MR. MÉNARD: Ten years of experience tell us that the ILEC aggregation network is not built to support really fast speeds and the only good way to solve this problem is to stop bundling aggregation with access and to let ISPs buy the service in a manner which is as much as possible unaggregated by ISPs bringing their own transport to central offices. We had this in the essential services proceeding and it is taking time to happen.
1588 THE CHAIRPERSON: I understand that but before we go to the unbundling, is there an up front investment required by the ILECs right now in order to achieve the solution that you want?
1589 MR. MÉNARD: Well, the ILECs have invested into buying -- you know, Bell, for instance, is known to have bought boxes from Arbor Networks, previously known as Ellakoya, and they have spent a lot of money for that and they have not increased their DSL wholesale rates for having purchased this hardware.
1590 So the point being is they have chosen to make an investment in a technology because they wanted to go do a job on behalf of the ISP. They wanted to open PPP and get inside the IP packet and do a job that really should have been done by the ISP itself.
1591 So I am not here to criticize their decision. I am simply making a statement observation that this role of traffic management ultimately belongs to the ISP, not to the incumbent carrier on behalf of the ISP. And if this technology was deployed pervasively, then there would be ways for incoming carriers to signal for the presence of congestion and the aggregation network and for us to react accordingly.
1592 THE CHAIRPERSON: But when you say if this technology was employed pervasively, what does that mean? Obviously it's not being employed right now. So, are you talking about a major technology outlay for the eyelet?
1593 MR. MÉNARD: Actually, that's the other way around. It's a major technology upgrade for the ISPs because the ISPs have not typically invested into traffic management hardware because it's extremely expensive, like the 4000 rotor front anagram is, you know, roughly $35,000.00 U.S. So, it's probably the most expensive rotor that an ISP can purchase.
1594 And the ISPs which have been given traffic shaping on a silver plater through unbundle -- through Gateway access service are, you know, some of them, members of the coalition, their position is that "geewizz" Bell made an investment that we didn't -- we would have needed to make and then, we don't need to make any more.
1595 And really, you know, as we look at this on a ten-year cycle, this is just one way our problem was solved, by one company for one wholesale service, it's not an industry issue and the difficulty I have is to reconcile the very good arguments of all the open internet coalitions and everybody on the application side with the reality that network operators have, which is that they need to engage into traffic management because the networks are not strong enough to support peer to peer applications today.
1596 THE CHAIRPERSON: Let me make sure I get this correct. Well, we're having our Next Generation hearing in November. What you're basically telling me, if we come to the right conclusion there, that will then force ISPs to acquire the necessary technology so that they, at their end, can in effect control their use by users in their demands on the network and, therefore, congestion of traffic and the corollary traffic shaking becomes unnecessary?
1597 Is that sort of and if you follow your train of thought, is that where you wind up?
1598 MR. MENARD: Yes, but that's only part of the issue. What I am actually hoping is that during the Next Generation hearing, the Commission will have -- having learned from this internet traffic management proceeding, will have learned the importance of costing wirespeed aggregation where the access network is built in such a fashion as the output speed is at least as fast as the sum of all inputs in order to alleviate any requirement for a traffic management to take place inside the access network and it's possible today, can give you the internet switches are inexpensive now.
1599 So, it's easy to aggregate at really really fast speeds and to send that aggregated output towards the ISPs.
1600 And if that is the focus and we can arrive at the true cost of that network architecture, there will be no need to implement IP traffic management inside the access network.
1601 THE CHAIRPERSON: Is this what you referred to in paragraph 44 when you say: CISP members are willing to work with the ILEC partners as their recognized traffic management for peer-to-peer areas in industry wide problem which will not go away, unless the unfairness of such application mitigates the flow management and by requiring the development of P2P application to work with ISPs and implement congestion control frameworks.
1602 MR. MENARD: That's in today's wholesale environment.
1603 THE CHAIRPERSON: That's today, into today.
1604 MR. MENARD: In the future wholesale environment that will be way simpler than to engage into doing that.
1605 THE CHAIRPERSON: Now, are you speaking for -- what about your sister organization, CAIP? I will be hearing them tomorrow. Are you both advocating the same thing or is this a specific solution that you're advocating?
1606 MR. MENARD: I believe that I am trying to reconcile what is the primary differences between what CISP members do and what CAIP members do and I think I have arrived at the conclusion that CISP members are doing quite systematically, all of us are doing cable modem wholesale rather than just ESL and so, we have a more general view of the problem, one which is not focused on the solution is it should be focused on DSL only.
1607 TPIA is a nice service, it does not quite work the way it should yet and -- but, this is where we are sending most of our business today on that --
1608 THE CHAIRPERSON: That is the distinction from CISP and CAIP, that you focus on TPIA and they focus on DSL?
1609 MR. MENARD: I am just saying that we do a lot of DSL too, but the point is if you're -- if I'm putting myself in the shoes of the Commission trying to reconcile well, why CISP has sometimes arguments that differ a little bit, it has to do with that as well as the fact that some CISP members are facilities based carriers, such as our company, for instance.
1610 THE CHAIRPERSON: O.k. Thank you. Len? Anyone of my colleagues have any questions? Go ahead. Allez-y.
1611 CONSEILLÈRE LAMARRE: Merci, monsieur le président.
1612 Monsieur Ménard, vous nous avez expliqué, là, de long en large les difficultés inhérentes au fait que vous avez à traiter, vous, en tant que fournisseur de service internet avec les titulaires qui fournissent les services de gros.
1613 Mais en tant que fournisseur de service internet, vous avez aussi des clients de détail et vous avez des responsabilités vis-à-vis ces clients de détail-là et vous-même, vous le reconnaissez là dans votre réplique du 30 avril, entre les paragraphes 26 et 27, là, si je peux vous citer, vous avez inscrit un titre qui dit: "ISPs are responsible for the end-user, not the wholesaler carriers".
1614 Bon, ceci étant dit, malgré ces responsabilités-là que vous reconnaissez, il n'y a rien dans vos documents qui fait référence à la question de la protection de la vie privée de vos clients.
1615 Et ce matin, suite à votre présentation, je reste sous l'impression que les membres de votre organisation soutiennent, à l'intérieur de certaines limites, l'utilisation de l'inspection avancée des paquets comme étant une méthode qui puisse être utilisée pour essayer de gérer la congestion.
1616 Et cette méthode-là, la critique principale qui est faite, c'est que c'est une méthode qui serait invasive au niveau de la vie privée de vos clients.
1617 Alors, j'aimerais ça vous entendre parler de ce que vos membres font pour protéger la vie privée de ses clients et aussi, si la technique de gestion du débit que... dont vous faites la promotion, si cette technique-là a des impacts, selon vous, sur la vie privée de vos clients?
1618 M. MÉNARD: Merci de la question. Donc, la protection de la vie privée des utilisateurs d'internet, c'est une réalité que tous doivent composer avec. Et ce que j'ai mentionné plus tôt, c'est que le seul moyen qui existe actuellement pour assurer la vie privée sur internet, c'est d'utiliser de façon libérale et pervasive l'encryption des données, de façon à ce que si le trafic est intercepté, *storé+, archivé, et caetera, que ce soit impossible de recouvrir les données privées qui sont à l'intérieur.
1619 Et, aujourd'hui, les techniques d'encryption sont très bonnes, mais ça prend... ça prend des ordinateurs immenses pour venir à bout de passer au travers de l'encryption de base qui est utilisée aujourd'hui. Donc, ça sera encore plus difficile avec les prochaines technologies d'encryption qui existent et qui vont être développées dans le futur.
1620 Donc, dans tout ça, ce que ça implique comme pattern, c'est que comme fournisseur de service, si on veut faire de la gestion de trafic qui est efficace, on ne peut pas se fier à l'intérieur de données qui sont encryptées. Il faut se fier au stamp sur l'enveloppe, pas au contenu de l'enveloppe.
1621 CONSEILLÈRE LAMARRE: O.k.
1622 M. MÉNARD: Il faut se fier à l'adresse qui est écrite sur l'enveloppe et non pas l'intérieur de l'enveloppe.
1623 Et il n'y a absolument aucune mention dans notre intervention de l'utilisation de * deep packet inspection + pour des fins de gestion de trafic. Le focus de notre intervention c'est de démontrer qu'il y a des façons de faire de la gestion de trafic en se fiant qu'à l'en-tête du paquet IP.
1624 Il y a plusieurs fournisseurs de service internet... nous, actuellement, même dans notre compagnie, on n'a pas encore déployé des technologies de gestion de trafic. On évalue la technologie d'anagramme et on va éventuellement déployer quelque chose, mais pour nous il serait mal venu d'argumenter les bienfaits de la gestion de trafic qui est faite de façon responsable en déployant des technologies de deep packet inspection.
1625 Donc, on s'est abstenu de le faire et je pense que la poursuite du dossier au niveau du Conseil va être très utile pour nous aider à voir vers où vous voulez aller.
1626 CONSEILLÈRE LAMARRE: Mais un des objectifs qu'on avait fixés pour cette audience-là c'était d'évaluer l'impact que les techniques utilisées peuvent avoir sur la vie privée des utilisateurs. Et si je comprends bien ce que vous dites, c'est que vos membres, présentement, n'ont pas nécessairement déployé de telles technologies, mais ils sont en train de l'évaluer.
1627 Dans cette évaluation-là que vous dites vouloir responsable, où est-ce que le critère de la vie privée de vos clients est introduit?
1628 M. MÉNARD: Le critère de la vie privée des clients est introduit au sens où la façon facile de faire de la gestion du trafic; c'est-à-dire de prendre avantage du fait qu'aujourd'hui les protocoles de type peer to peer ne sont pas tous encryptés, permet à certain carriers aujourd'hui de réagir de façon utile, dans le sens où ce n'est pas totalement useless les technologies de DPI actuellement parce que BitTorrent et autres sont déployés at large sans encryption.
1629 Ce que j'essaie de faire valoir comme point, c'est que ça va arrêter un jour parce que la conséquence de tout ça, c'est qu'on va... tout le monde va devoir se mettre à traiter toute forme de trafic encrypté comme du mauvais trafic puis de ralentir systématiquement dès qu'il y a de l'encryption. Et certains carriers font ça aujourd'hui.
1630 Donc, la protection de la vie privée, elle va toujours être protégée si elle est encryptée. C'est juste que ça va marcher comme de la... ça va marcher tout croche.
1631 CONSEILLÈRE LAMARRE: Oui, mais êtes-vous en train de me dire que si elle n'est pas encryptée, vous estimez que le distributeur ou le carrier, à ce moment-là, a la liberté d'inspecter ce qui se passe sur ses connections?
1632 M. MÉNARD: On ne prend pas position sur ce point-là parce que ne pas encrypter sur internet, c'est une invitation à ce que quelqu'un le fasse essentiellement. Donc, d'une certaine façon, c'est ça la réalité d'internet aujourd'hui. On ne peut pas...
1633 CONSEILLÈRE LAMARRE: Oui, mais, moi, je parle de la responsabilité, là, des distributeurs.
1634 M. MÉNARD: Oui.
1635 CONSEILLÈRE LAMARRE: Vous, vous avez une responsabilité vis-à-vis de vos clients.
1636 M. MÉNARD: Oui.
1637 CONSEILLÈRE LAMARRE: Alors, je suis d'accord avec vous que c'est une invitation, mais c'est une invitation en ce sens que c'est une invitation qui... bien, de qualifier ça d'invitation, c'est un peu pervers. C'est comme de dire que d'avoir une voiture et de la mettre dans un stationnement public, c'est une invitation à se la faire voler.
1638 M. MÉNARD: Si tu laisses les clés dans l'ignition puis les portes ouvertes, il y a des bonnes chances que ça se fasse.
1639 CONSEILLÈRE LAMARRE: Sauf que ça reste un vol quand même.
1640 M. MÉNARD: Absolument.
1641 CONSEILLÈRE LAMARRE: Bon. Alors, ceci étant dit, et je comprends que vous ne voulez pas prendre position, mais ce que je veux savoir présentement avec ce qui se fait, mais ce que je veux savoir c'est si, lorsque vos membres vont déployer des technologies de gestion de trafic, si la protection de la vie privée de leurs clients, de vos clients, ça va être ou dans la liste des considérations et des priorités et comment est-ce que vous estimez que vous allez pouvoir la protéger?
1642 Et est-ce que la gestion de débit dont vous faites l'apologie, là, dans votre présentation, est-ce que c'est un type de gestion du trafic et de la congestion qui permet de protéger de façon efficace, selon vous, la vie privée de vos clients?
1643 M. MÉNARD: Absolument, parce qu'on ne prend pas de décision de gestion de trafic basée sur le contenu d'un paquet, sur une application donnée. Donc, nous, tout ce qu'on regarde c'est si l'utilisation générale que fait un usager de sa connexion de vitesse cause une congestion sur le réseau et, à ce moment-là, là, où on va agir, c'est en disant aux applications et habituellement c'est au système d'exploitation des ordinateurs de ralentir.
1644 Et c'est la responsabilité des systèmes d'exploitation des ordinateurs de faire le relais de l'information par l'entremise de la pile TCPIP, là.
1645 CONSEILLÈRE LAMARRE: Oui.
1646 M. MÉNARD: Le mécanisme TCPIP de dire à l'application: aie! tu as perdu un paquet, ralentis. Donc, en fait, le système d'exploitation va le faire au nom de l'application parce que c'est le système d'exploitation qui gère la gestion de la congestion.
1647 Donc, ça ne peut pas être mieux. C'est ça essentiellement qu'on veut faire valoir. C'est qu'il y a des façons de faire des gestions de trafic qui sont entièrement responsables de la vie privée des usagers et qui ne vont pas décourager l'utilisation d'encryption.
1648 Ce qu'on pourrait faire qui serait vraiment néfaste pour le futur de l'internet, c'est décourager l'utilisation d'encryption en donnant un accès prioritaire au réseau, aux données qui ne sont pas encryptées.
1649 CONSEILLÈRE LAMARRE: Merci. Là, je comprends mieux votre position. C'est tout, monsieur le président.
1650 THE CHAIRPERSON: Okay.
1651 COMMISIONER DENTON: Mr. Menard, I just want to go over the -- but I think it's the essence of your arguments too whether you agree, whether I have understood it correctly.
1652 First, you are saying that encryption will shortly void the value of deep packet inspection.
1653 MR. MENARD: Yes.
1654 COMMISSIONER DENTON: Secondly, you are saying that better signalling of congestion is needed?
1655 MR. MENARD: Yes.
1656 COMMISSIONER DENTON: Third, you are saying this function, the signalling of congestion should be done by the ISP?
1657 MR. MENARD: With the help of the incumbent carrier, which is to advise that it's never congested.
1658 COMMISSIONER DENTON: Thank you. And doing this at the carrier level, doing the -- engages doing this, what I mean by this is doing the inspection at the carrier level engages them in too much packet inspection?
1659 Do they have to use too much intelligence to get into -- computer intelligence to get into the signal if they do it at that level?
1660 MR. MENARD: The point is that they should be statutorily foreclosed to act on my behalf as an ISP and to start messing up with my end-users IP traffic.
1661 COMMISSIONER DENTON: Okay. And the first part of the solution then is that better physical and business arrangement of the aggregation among ISPs and carriers will largely obviate, that is to say eliminate the need for traffic management?
1662 MR. MENARD: That is correct.
1663 COMMISSIONER DENTON: And then, in the meantime, better network congestion signalling mechanisms are required?
1664 MR. MENARD: Absolutely.
1665 COMMISSIONER DENTON: Thank you.
1666 THE CHAIRPERSON: And now the only thing right now, they do not have the necessary signalling message around. That system does not -- right?
1667 MR. MENARD: There is no signal.
1668 THE CHAIRPERSON: There is no signal. Okay. Thank you very much. Those are our questions.
1669 Madame la secrétaire, je crois qu'on peut prendre la pause du midi maintenant? À quelle heure est-ce qu'on va recommencer?
1670 LA SECRÉTAIRE: Nous allons reprendre à 13 h 15.
1671 LE PRÉSIDENT: Merci.
--- Suspension à 1141
--- Reprise à 1319
1672 LE PRÉSIDENT : Bon, commençons, Madame la Secrétaire.
1673 LA SECRÉTAIRE : Merci, Monsieur le Président.
1674 I would now invite the next participants, Mr. Jason Roks, Mr. Norm Friesen et Vaxination Informatique to come to the presentation table as a panel.
1675 It appears Mr. --
1676 THE CHAIRPERSON: You're one in three; are you?
1677 MR. MEZEI: Pardon?
1678 THE CHAIRPERSON: You're one in three or three in one?
1679 MR. MEZEI: I don't think so.
1680 THE CHAIRPERSON: All right.
1681 THE SECRETARY: Mr. Roks and Mr. Friesen don't appear to be in the room.
1682 We will then proceed with the presentation by Vaxination Informatique. Appearing for Vaxination informatique is Mr. Mezei.
1683 You may begin your presentation.
1684 M. MEZEI : Jean-François Mezei de Vaxination Informatique.
1685 I want to thank you for this opportunity to present. I'm just a citizen, I'm a self-employed consultant and I followed these issues when they started because I was very concerned as a citizen and I felt I could contribute to this and I'm here today hoping to help you with this.
1686 Last month the Prime Minister of Great Britain went on a project called The Digital Britain because they realized that they were falling behind and they wanted to catch up. And he said in the Times of London, the Internet is as vital as water and gas.
1687 And I think this is something that's very important to consider with all that's going on right now because a lot of countries, and Canada included, start to rely on the Internet, not so much because of a dropped packet here or there or congestion, but you rely on it to file your tax returns, to buy airline tickets, if you don't do it on the Internet you have to pay $20 extra, so on and so forth.
1688 And it has to become a utility that people can rely on, and a utility doesn't care -- whether you use electricity to power a hair dryer or your toaster, you buy electricity or you buy -- you pay for your water amount for a year and you use it in a reasonable way obviously, but you use it the way you want and the supplier doesn't care what you do with it.
1689 Another important point, and you will see why I come to this later on, but to be competitive on a global basis a country needs to have an efficient telecommunications infrastructure and I think you've listened to Mr. -- from the Zip.ca this morning who said that hosting his servers in Canada was more expensive, the bandwidth was more expensive in Canada and a lot of people are hosting it in the States.
1690 And essentially, basically if we don't have a competitive environment with advanced leadership role in telecommunications, we risk seeing all our brains and our businesses move to other countries and at that point Canadians become dependent on other countries, and that's what I call loss of sovereignty.
1691 So I think, you know, we're dealing with these issues here that don't seem on the surface so important, but they are because they set a direction and they set a tone.
1692 I want to talk to you about one of the aspects that got me involved was hearing all the speeches about how bad P2P or BitTorrent was, and I've seen a lot of speeches from the telecommunication companies and a lot of corrections need to be made, you'll see about this.
1693 And obviously there's a discrimination aspect that's been talked about because of the fact that DPI, the way that it works to make decisions, there is discrimination aspects.
1694 And, obviously, the word innovation has been mentioned a lot and I will talk about this point -- my point of view, competition, and finally regulation.
1695 First of all, you've all seen these famous graphs that multiple carriers have presented to you, Bell, Rogers and I think some of the DPI manufacturers have done this, where they claim that because peer-to-peer uses many TCP connections that they can use up to 11 times more bandwidth than someone who uses one.
1696 First of all, when you've got a five megabit per second link to your house you can't use 11 times more bandwidth than your neighbour because you're limited at five megabits per second, and I think that the people who've shown you these graphics, it's not been very, very helpful for them to not mention that.
1697 There is one aspect where this is actually true is when you have two servers in the lab, like a university lab for instance with 100 megabits between the two, this will actually happen where your P2P application that uses, let's say, 10 TCP connections will actually take more bandwidth than your lone application.
1698 And this is because of the way the software is structured, and this comes back with what Francois Ménard said, you get a nice little slide here.
1699 Basically your software -- you have your application, either P2P or mail or http, the web, they're at the top and when they connect to something on Internet they will talk to either the TCP stack or UDP which was mentioned -- you can ask me questions if you want, clarifications later -- that one actually then talks to the IP stack and then the ethernet stack to get to the next device, your modem, and the modem then drops the Internet and uses another ADSL to get on.
1700 And all TCP applications behave the same way on the Internet and the packets at the header level -- and this is very important -- the packets are the same at the header level, mail or http or the web or P2P, the packet headers have the exact same functions or the same properties in them, the same bits and the same types of service and the same flow control.
1701 And, so, for the Telcos to actually go out and say that P2P uses a lot more than normal -- other applications, this is very misleading because they all use the same core software and that's that software, TCP, which is deployed around the world, has been running for, I don't know how many years, since the early 80s if not before in wide deployment, it's the one that has that.
1702 The one thing I want to mention about IP, and this may explain a bit about UDP later, UDP is really IP with a few extra bits added.
1703 IP is basically just shouting, you shout -- you send a packet from A to B, you address it but you don't get a reply saying yes it's received, you don't even know if it's been delivered or not, so it's an unreliable method of doing this.
1704 TCP gives you the reliability because you wait for an acknowledgement back that the packet's been received and the acknowledgement -- the time that it takes for the acknowledgement to come back allows a TCP stack to measure the type of performance the line has.
1705 And when you drop -- when someone drops packets, either intentionally or due to congestion, the throughput isn't actually reduced, that's just a side effect, it's just a TCP routine that decides I'm going not going to send as many packets before I get an acknowledgement and that ends up resulting in slower throughput. That's the side effect.
1706 Okay. The next one now. You have here two computers, one running the web and one running the P2P applications. The P2P may have 10 TCP links but they're limited to five megabits per second, and the web may have one TCP link and it's going to take the five megabits at once.
1707 From a network point of view, once it gets into the IP network they're just IP packets. The network itself looks at the IP header and says this is destined for that IP address, I will take it and send it there.
1708 It's not aware of TCP sessions, the TCP session itself with the -- Telcos call the flows, is something that's understood only by the two end points. They know what their TCP parameters are, everything in between is unaware of this.
1709 The networks don't need to know about the flows to manage their network. They like to know because they like to -- the word spy is not right -- but they like to have statistics on what their users do and that's -- in my opinion, it's a privacy issue because they've got no business knowing, you know, if my http session lasts five seconds or 15 seconds or if, you know, my mail is a long session or not. What they care about is the number of packets.
1710 And from the point of a view of a network, when you send it five megabits per second either with 10 TCP sessions or a hundred TCP sessions or one TCP session, it's the same number of packets and the same throughput on the network.
1711 So, the networks have no business saying that it takes up more bandwidth than other applications.
1712 And I'll give you a very good example of this. Basically I did tests last year during the CAIP proceedings. I downloaded a movie through peer-to-peer and it took a hell of a lot longer because of the throttling and I downloaded a movie from the Bell Video Store, which is now defunct, and that went full five megabits constantly.
1713 So, the Bell Video Store uses the same amount of bandwidth and creates the same amount of congestion because it's just a TCP session it uses. As has been said by the Telcos, it uses as much as is available, which is what the network is designed to do, there's nothing wrong with that.
1714 Okay. And this, again, YouTube is an interesting phenomena because it's now the biggest user of bandwidth on the Internet and they're getting into HD movies.
1715 And I'll give you an example. In early June there was a movie from a French producer called "Home" which was distributed for one day for free around the world, it was on YouTube in high definition, an hour and a half, downloaded non-stop and this would have taken -- it was also available on peer-to-peer networks and it would not have made a difference whether it's downloaded from YouTube or peer-to-peer.
1716 In fact, peer-to-peer, one can argue it's more efficient because it comes from different places to you which means that to the ISP, if he has five different links to the Internet the load will be distributed on the five links, when it comes from YouTube it comes over one link and that will over saturate that one link, but that's at the network level.
1717 The irony of this right now is because the networks have chosen to throttle one application you can download a five gig HD movie from iTunes or from YouTube unthrottled, but you want to download a 300 meg or a 30 meg piece of movie or pictures or animation over peer-to-peer and it's throttled down to, you know, stone age speeds basically.
1718 And that's really unfair because, if you're looking at congestion you're not -- you should be looking at who's using the bandwidth as opposed to what application because different applications can use more or less bandwidth, it depends on what you do with them, not the nature of the application itself.
1719 You know, as I said, you can download a big piece or a small piece either with the web or with peer-to-peer and these two applications behave exactly the same way.
1720 So, it's unfair to allow a network to discriminate between the two because from the congestion point of view, there is no difference.
1721 Okay. Now, discrimination's been brought up and what's important here for the Commission to admit, which it has not done so far, DPI equipment looks at packet contents and you have to look -- this is fully documented on the Internet, all the network protocols are documented, there are standards, the packet header is from byte one to byte "x" and after that is the payload, and if the DPI equipment looks after the end of the header it looks into the payload.
1722 And the Commission, the country, the nation has to say, yes, DPI looks at the content, whether what it does with it is another question, but you have to admit it looks at the content.
1723 And in my presentation -- in my submissions last year in the CAIP file I provided all the layouts of the record formats and stuff and show that from the headers only you cannot differentiate between P2P or http or another.
1724 And that's important to know and important to note, because they use DPI to guess what you're doing. They look at bytes in the payload and they're trying to guess what you're doing and there are errors.
1725 Bell Canada basically knows if it's an encrypted connection and by default it will throttle encrypted unless it's a specific one, a VPN one on a specific port, and that was admitted by Bell I think last December or something. They had problems with secure FTP which is being throttled even though it shouldn't.
1726 And that's why this is -- I underline guess and deduct because this is not an exact science. And Mr. Ménard this morning also mentioned, they have to subscribe to get updates because they change.
1727 And again, network management should focus on bandwidth used, not the type of application.
1728 Now, we have here peer-to-peer and this is where I get maybe a bit of black ops or tin foil hat, but it needs to be said. YouTube is established, it's mainstream, it's talked about in the media, people use it a lot and there's no way that an ISP could throttle it because that would backfire.
1729 Peer-to-peer is still at the infant stage used by early adopters. In Canada it does not yet have commercial applications although it does in the U.S. and which is one of the reasons Comcast was told, stop throttling P2P because you're hurting applications -- commercial applications.
1730 Basically, because it's an emerging technology that has the potential to transmit lots of data, the feeling I have is the big legacy Telcos don't want it to take root, so they're throttling it now before it becomes popular. They didn't do this for YouTube and now it's too late, YouTube is sending out HD content unthrottled and unmanaged.
1731 And I think this is something that the Commission has to take grasp at because in the long term you can't afford to allow the Telcos to start control -- deciding what to nip in the bud and what to allow, and that comes into the innovation aspect.
1732 One last thing, P2P is democratic. People can start their own streams, you can create your own content and distribute it yourself, there is no control over how it's distributed; whereas if you look at iTunes or Google or YouTube, you have a central body here that controls the distribution.
1733 And I'll give you an example. By the way, those graphs that the Telcos use that I showed early, they came from a study done by British Telecom and the purpose of the study for British Telecom was to prove that the BBC was generating too much content and British Telecom wanted to get paid by the BBC and that's because they could focus on one company, the BBC, send the bill.
1734 P2P because it's democratic, anybody can start a BitTorrent, the Telcos can't have any control over it and I think they're sort of concerned about it and that's one of the reasons they decide to pick on that.
1735 THE SECRETARY: Excuse me, Mr. Mezei, your time is almost up. Can you conclude, please.
1736 MR. MEZEI: Okay. I will quickly.
1737 Based on yesterday, I can talk about this, it was asked yesterday whether there are solutions without DPI. There is a small ISP in Montreal called VIF, they have had a solution for 12 years where they actually reduce the speed of users once they have exceeded their monthly allotment and it's reset after a week or something.
1738 Very quickly, the big thing about congestion is because the telcos are increasing the speeds and when you increase the modem speed you enable new applications. If we stayed at 300 baud we would not have the ability to download movies or even music, but the telcos compete on the speed only and they raise the speed like they want without being able to actually provide the bandwidth below that. That's where there is congestion.
1739 But if the Commission were to say "Look, you need to tell the people how much speed you can sustain. Not just burst at 25 Mb for 2 seconds, you can download for an hour at 30 kB per second", and that should be in their advertising. If the telcos are forced to advertise what their actual capacity is given to each user, they will compete on that and they will quickly upgrade their capacity because that's what they compete on, instead of fake speeds that nobody can reach anymore because the minute you start downloading something big they throttle you.
1740 I guess my time is up. I will be open for questionning I guess.
1741 The process now goes to him; right?
1742 THE SECRETARY: Yes.
1743 Mr. Jason Roks has now arrived, so we will proceed with Mr. Roks presentation and then followed by panel questions.
1744 You can start anytime.
1745 MR. ROKS: Hello and thank you for inviting me to this hearing. I appreciate making the effort to have me come out.
1746 I think everybody has my set of slides. I'm not putting it up on the screen, I don't think there is any need for it. Yes? Okay, excellent.
1747 Just quickly on myself, I have been affiliated with the CFC Media Lab for years, one of the founders of Wireless Toronto. I have been involved with Save the Internet and Free Press to help get word out about what's going on in the industry. I helped launch SaveOurNet.ca and on my own decided to put my own submission in to cover other issues that aren't particularly being covered by the other parties that are presenting.
1748 As a result, it also turns out that I'm working on a Ph.D. on this particular topic and the results will come out after these hearings in this country and in other countries and I will be analyzing the fallout from what the decisions are made.
1749 My background actually comes back from peer-to-peer networking -- or what was called peer-to-peer networking back in a company called Hotline. We had 14 million users using -- it was our own protocol, it was back in 1996, and we were facing a lot of the same issues where people were saying "Well, what's going on with your protocol?"
1750 It was very fast. If that protocol was still around today and still being used, then it would be providing the same kind of effects on the network as Torrent or other peer-to-peer applications are doing, and it wasn't even peer-to-peer, it was like FTP.
1751 I spent some time at the CBC working on their peer network infrastructure across Canada which helped them save a substantial amount of money on delivering media and data across the country at a fixed cost. I will also get into that as a potential solution to the ISPs for the problems that they are facing.
1752 I also set up the worldwide network infrastructure for the Real News Network, which is an independent news organization, analyzed Corus, worked with Times of India, Zoom. I currently, out of disclosure, do some work for Rogers at the moment as an advisor and I also run a search company called GUIgoog.
1753 I'm not really sure why I'm here, but I appreciate that I am here. My submission covered four main topics, peer-to-peer and Torrent, to demystify -- I'm sorry, five topics -- number two, the network and network upgrades; number three, BBU versus BBS, meaning billing by usage versus billing by speed; peering and Internet exchanges and the mobile Internet.
1754 I will just basically go over them again. My submission is on the site, if you can find it. It's in the digitally submitted.
1755 So I will start with peer-to-peer and BitTorrent.
1756 A lot of my concerns is the confusion and myths that circulate about a lot of the information about these hearings.
1757 Number one, Torrent file is just like a package or a box and you put things inside it, whether you put a video in it, an image, it's just like shipping container, shipping it from one location to another. The idea that one would be blocked just because of the shape of the box is a bit concerning to me. We don't do that in the Postal Service.
1758 There are two parts when discussing BitTorrent.
1759 Number one, the file, which is a small little 10K file which is metadata. It describes the file. The file actually doesn't really exist anywhere, it exists in pieces all over the place.
1760 The other part of BitTorrent is the protocol, the protocol being what controls how the bits are sent back and forth.
1761 So I'm just trying to get a clear -- clarify an understanding of what Torrent is. It is essentially the same as any other file format or protocol out there and there will be more to come.
1762 Although it's not formally recognized by the W3C -- I may be wrong, but I'm pretty sure it's not recognized yet -- BitTorrent was built for efficiency and it's an emerging technology. To ban, hinder, block a file type or application is a measure that stifles an emerging technology and it wouldn't stop here. If this sets precedent that a technology can be banned, blocked, hindered, that's just setting us up for stifling innovation.
1763 BitTorrent just happens to be what people are using now, it used to be Kazaa, Scour, Napster, FTP, IRC, newsgroups, Hotline, I can go on. And BitTorrent might be what's in use today, but there is going to be a new one tomorrow and people's need or desire to consume data, media, entertainment, whatever they are doing that is causing this, is not going to lessen. It's not going to go away.
1764 Throttling doesn't solve the problem. The real problem is not the peer-to-peer traffic because the traffic, as I said, will never go away and it's becoming adopted into mainstream applications.
1765 Skype, which most of the world -- it's actually I think is the largest Internet phone call company in the world right now, they use peer-to-peer technology and that's voice over IP. Is that going to get blocked? Is that going to get throttled?
1766 Adobe flash video 10 incorporates -- I think it's a technology called Octoshape. I'm not sure, it's octo something out of -- I think Estonia is where it's from. That is peer-to-peer technology.
1767 And in fact Obama's recent video was done in that technology.
1768 And today I just got word that there is 18 Gb per second running over the local loop in Toronto for the Michael Jackson memorial or funeral or whatever it is today. That's actually a record. All that data is free data. Keep in mind. I will get to that point later.
1769 High bandwidth applications aren't going to go either. They are not going away. Ironically it's encouraged by ISPs to play more games, watch more videos, make more video calls. So on one side they are saying don't use it, but on the other side they are marketing to the public saying "Here's some great speed, use it; use it more, use it for faster things."
1770 Torrent protocol also supports encryption and currently deep packet inspection does not see encryption. So if that is what DPI is meant to try and manage, it's ineffectual because it's encrypted data, it can't be seen. So it only works on things that aren't encrypted.
1771 The only way to watch that would be to watch a port to see if there was an odd usage of ports out there, like for example the Web uses port 80, secure web uses port 81, FTP uses port 21, but if I, just for example, put my BitTorrent port to port 81, the IFP would just think it's Web traffic secured.
1772 So there is no way to really stop it and the fact is it's not going to go away.
1773 I bring this up because it means you either adopt a new technology or do you push it and stifle it?
1774 There is a company in Israel called Bezeq International, took the opposite approach adding local nodes and trackers for BitTorrent inside their own UserLand for the subscribers. In fact, what it ended up doing is speeding up -- or that's the objective, is to speed up the Torrent download to get them out of the way.
1775 Congestion is when something ends up in a queue and if things are in a queue you need to speak that queue up, not just eliminate the things in the queue. I will just leave it at that.
1776 And I would say that the problem is the network and that is what I think the Commission should remain focused on and as solutions to building out the network required to sustain the demands of Canadians.
1777 The network, many people point the finger at the ISPs saying the problem is their own for not upgrading their network.
1778 It's also being pointed out that the congestion is happening in between the plant and the end-user, which is also known as the last mile. This is on the ISPs network. This is not effect of networks outside of their control, this is within their own network infrastructure. So the problem is with the ISP and their network capacity.
1779 The response you get from ISPs is actually usually the same, "That would cost a lot of money". And then you ask them "How much money would that cost?" And they say "Oh, that's a lot of money." "Well, how much money is that going to cost"? "That's a lot of money."
1780 So we don't have any numbers, as far as I know. From all the materials that were provided I didn't see any numbers on what it would cost to upgrade the network, so I decided to go around and dig and ask a few people some questions -- how am I doing for time?
1781 THE SECRETARY: You have two minutes remaining.
1782 MR. ROKS: Okay.
1783 So I looked into it. An ISP of 2 million subscribers has approximately 130 plants which are nodes that are required to service the area. An upgrade of line cards at the plant level are required and likely software and training to manage that. That will help deal with the congestion issues, upgrading the actual network.
1784 The hardware costs alone are $100,000 a card. Wow, that's a lot of money. And then along with software and training I will double it to $200,000. So at $200,000 you have an average plant that requires five cards, each providing 37 Mb capacity, 185 Mb total, approximately $1 million per plant.
1785 So the total cost for someone with 2000 subscribers to upgrade their network to try to deal with today's bandwidth loads is approximately a $130 million upgrade at my inflated prices.
1786 But let's say they were accurate, that's $65 per subscriber. Amortize that over three years and you have $21.67 per subscriber per year. Add $2.00 a month for the next three years.
1787 BBU versus BBS. Billing by uses versus billing by speed.
1788 The Internet has always been billed by speed. The data rate that one connects to the Internet, there's upstream and downstream, which they are sometimes different, but usually refers to both.
1789 There have been many speeds, you have probably all heard of them, 19, 228, 833, 6, those were your dial-up days. You have your 28-8 and your 256, which is your high-speed lite, your 1 Mb, which is your high-speed, and then you have your 1 megabit plus and your super high speed, your extreme, all these marketing ploys to sell people speed and use more data, yet they turn around and say we have too much data being used. It is a conflict and it's problematic for people who are trying to get and understand this service that they are purchasing.
1790 Billing by speed actually makes it easier -- actually billing by speed has always been the way the Internet has operated. You used to have a 33-6 modem, a 28A and a 14-4 and it was always billed differently by speed. The speed in fact was the cap. If I could only do 1 megabit per second my cap is my speed. I can only reach 1 megabit per second. If you extrapolate that over day -- I don't think my number is right in here, I didn't correct it, but it says 4 GB a day, I think it's actually less than that, but that is my cap.
1791 But then to have an ISP turn around and say "Now I'm going to bill you on the gigabytes you're using", it's "Wait a second, I have already paid for those gigabytes. I paid for what the throughput of my connection is. If you can do 1 megabit it should do 1 megabit for me. I shouldn't have to pay for using the 1 megabit that I'm paying for." I hope that makes sense.
1792 Overselling. That seems to be the problem that I have identified. As a repercussion of using speed as a marketing variable, ISPs have brought congestion upon themselves by allocating too much bandwidth per end-user based on the capacity of their current networks.
1793 An average user, actually surprisingly, has a sustained usage of 70 K a second. That's less than an audio stream, and a low-quality audio stream at that. One megabit is what's required to consume most and to do most things on the Internet except HD video. So watching YouTube, you can pretty much do everything in one megabit. It's plenty enough speed to do.
1794 And by the average usage we see that that 1 megabit need is a burstable need. It's I only need it sometimes, I don't need it all the time.
1795 But if all of a sudden I had a 10 Mb connection and my neighbour had a 10 Mb connection and he has a 10 Mb connection, now we are all trying to -- we are all accessing a limited amount of space with too much -- I will leave it at that until I get to the actual point.
1796 So there are two types of data, transit data and local data, which is also known as peer data.
1797 A lot of people don't acknowledge the different types of bandwidth out there.
1798 The difference between peer bandwidth and transit data is transit data is when a request goes outside the network of an ISP. They have to then connect to another network, which costs them money to connect to that next network. That's called transit data.
1799 Free data is the data that goes amongst their own nodes that isn't going outside to another network. That's called peer data, loop data, local looping data, free data is also what it's known as.
1800 Peering actually creates efficiencies that help deal with local congestion issues. It in fact speeds up and lowers latency. So peering has been adopted by ISPs to manage local traffic and costs. So it reduces the cost of the bandwidth when you peer. As I said, it also reduces latency, it clears the queues faster and it maintains geographic privacy.
1801 So, for example, there is only one company that I know of in Canada refuses to peer and that is Bell Canada. Rogers peers, TELUS does some peering, Shaw peers, Cogeco peers. It's in the best interest of everybody to peer, but Bell refuses to peer. It would be in the best interest of everybody and put data through the network more efficiently and faster if Bell peered, even if it was conditional.
1802 Everybody else has conditional peering, I think Bell should step up and join the actual network and how the Internet was designed. By not peering they are breaking the fundamental rules of how the Internet works, which is sharing data.
1803 Wireless Internet. I will make it very quick. You have a phone on one side, you have a mainframe computer on the other, keep going and you end up in the middle at an iPhone. Wireless Internet is the Internet, it's all the same and we shouldn't even be talking about it.
1804 So I wanted to clarify whether the Commission means by mobile wireless that it is going to be governed by the same rules and guidelines that come out of this hearing.
1805 I answered your questions very quickly, I am getting to the end of my presentation. Essentially it's very clear, any traffic -- so question number one, accessibility of ITMP. Aside from security, spam or sentinels, which are worms, Trojan, harmful affects to the network being done maliciously, there should be no additional traffic management and should only be used in an emergency circumstance if there is congestion on the network, but as well only while the network upgrades are taking place, not as a replacement to avoid network upgrades.
1806 This set of guidelines should be developed by a consensus of engineers, network specialists, industry, public and academic.
1807 Service provider disclosure. Fully disclosed. Publish update on their website to the public, both to the consumer and the customer. I don't see any other full disclosure and transparency. The only exception would be for spam, security and sentinel.
1808 Privacy. Well, I mean, as far as I read recently, the Privacy Commission seems to believe that DPI is a privacy issue, so it really doesn't matter what I have to say and I think I'm just waiting to see what they have to say. So I will defer my comments for now.
1809 Strictly enforce guidelines required to contribute to the protection of personal privacy.
1810 And ISPs should be able to manage their network with clarity and transparency to the end-user.
1811 Wholesale services. Yes, it should also be included for wholesale services.
1812 And mobile carriers, that's the Internet.
1813 Finally, clarity of jurisdiction. Well, again is DPI a privacy issue? I'm not sure, but I think there's a Member's Bill in Parliament, there is a lot of talk about putting a Bill into Parliament.
1814 You have decisions that are being made on questions that are being asked in this hearing.
1815 You have telecom versus Broadcast Act. Do you merge them? Do you start an Internet Act or a Network Act? I don't know, but I mean those lines are being crossed. When we are talking about convergence of media, we are also looking at convergence of media.
1816 And is the Internet an essential service is a question I think we need to ask ourselves, especially when we look at government services, especially ehealth starting to move onto that platform. At that point does it become an essential service? I'm wondering.
1817 My recommendations. These are the same recommendations I have in my document. I'm out of time so I don't really need to read them, but my recommendations are all there on alternatives that can be used to try and find a resolution and deal with this problem, because the problem is not going to go away.
1818 Thank you very much. I appreciate it.
1819 THE CHAIRPERSON: Thank you for your presentations.
1820 Mr. Roks, I am confused. I have heard now from various people that P2P is a guzzler of the Internet, uses up five times as -- uses a huge amount of space, et cetera, and the issue actually is not only on downloading but on uploading and various networks have different capacity, the uploading capacity being much more limited than the downloading and therefore if you use P2P, you know, you cause congestion for the whole, et cetera.
1821 Yet you here in your presentation on slide 7 basically suggests that P2P behaves like any other application.
1822 Resolve this conflict for me. I mean are you -- obviously neither you nor the others are lying, it is just is I guess a different focus or different way of looking at it or different representation, but how do you reconcile this?
1823 MR. MEZEI: Actually, one has to look -- the way an application works is one thing and what the application is being used for now as the other aspect.
1824 When I say that P2P doesn't behave differently than other TCP applications, this is in terms of how it works at the network level.
1825 Right now what's happening is you have the early adopters of P2P who have discovered that with currently the new speeds that we have in Canada, which are nowhere near world leadership, but even at 5 MB you can start you download full-fledged movies.
1826 THE CHAIRPERSON: Yes.
1827 MR. MEZEI: P2P happens to be the technology that makes it easy and efficient to do. And so what you are seeing right now is those early adopters are using P2P to download the big movies and this is why you get the telcos presenting last year and the year before saying P2P is horrible and it's the end of the world because it represents that much.
1828 But since then YouTube has overtaken them in terms of the bandwidth used and there are going to be other applications.
1829 ITunes from Apple now lets you download HD movies and movies as well. That, when you download a movie from Apple iTunes, will take up as much bandwidth and create as much congestion -- because it'is also a TCP application -- as P2P.
1830 So what's happening now is the early adopters of P2P per se are now going to be complemented by a mass use of YouTube and iTunes and any other systems that come along that sells Mr. -- from Zip.ca this morning who wants to distribute his movies online -- the same issue.
1831 You have a certain amount of data to transmit and that's going to take bandwidth and it's going to use whatever bandwidth is available to move as quickly as possible. That's what the Internet was made for.
1832 So there is no conflict between the two, the fact is that what the telcos have presented to you last year and recently were the previous statistics, without looking ahead saying "Well, if we throttle P2P something else" -- as Mr. Roks says, "something else comes along". And something else has already come along, it's called iTunes and YouTube and everything else.
1833 So focusing just on P2P will not solve their problems.
1834 THE CHAIRPERSON: But here at this very hearing --
1835 MR. MEZEI: Yes...?
1836 THE CHAIRPERSON: -- yesterday somebody showed me a chart that 5 percent of the users use up 60 or 70 percent of the --
1837 MR. MEZEI: Yes. Early adopters.
1838 THE CHAIRPERSON: And they are all P2P users. And you are telling me that it just happens to be P2P, there may be some other technology tomorrow?
1839 MR. MEZEI: Yes. And the statistics they produce is basically early on they were the first one to start downloading movies and other media intensive content, so they were the ones who got noticed. And you are always going to have a 5 percent of early adopters whatever technology it is.
1840 This morning the mention of UDP was made. I would just like to point out, you have mentioned that P2P now uses encryption -- encryption is throttled by Bell, by the way, when you use P2P, because they know that it's available.
1841 The newest versions of BitTorrent now use UDP to bypass TCP completely and they have their own sort of mechanism for flow control which is not compatible with TCP.
1842 And why they are doing this, they are trying to evade the throttling. So the very nature of throttling will cause inefficiencies because applications will steer away from the most efficient path because that most efficient path has been throttled.
1843 THE CHAIRPERSON: I understand both from you and your colleague that, in effect, encryption makes application-based throttling very difficult anyway, or counterproductive, because you can encrypt so that -- whether they use DPI, or whatever technology, cannot detect what kind of application it is.
1844 MR. MEZEI: Yes, so they assume that you are doing -- if they can't detect it, they assume you are doing P2P and they throttle you.
1845 This is what Bell does.
1846 THE CHAIRPERSON: Now, you and the people we heard this morning all suggested that rather than going application-based, you should have a usage-based system, and, in effect, either through a higher rate or -- like VIF, which you mentioned --
1847 MR. MEZEI: Yes.
1848 THE CHAIRPERSON: -- have a maximum limit, and if you exceed that you get throttled, et cetera.
1849 Is the technology there? Is it available?
1850 MR. MEZEI: Yes.
1851 THE CHAIRPERSON: Why are they not doing it?
1852 And, please, be concise in your answer.
1853 MR. MEZEI: In the case of VIF -- and I actually found this -- I knew because people complained about VIF a few years ago, when they implemented it, and I knew about it, and I did some research, and I actually called them, and they confirmed that they have had this for 12 years, and they are using standard routers to do this.
1854 They don't need DPI equipment or anything fancy, they do the calculation of the bandwidth use using stuff that has existed for years, and when they detect at night that the user has gone over his limit, they program the router for that user, and when he logs in the next day, he is given an IP address that has low priority on the network and he gets throttled.
1855 THE CHAIRPERSON: But, in the meantime -- that is after the fact. In the meantime, if people use too much, don't you get latency or jitters or whatever?
1856 MR. MEZEI: Yes, but companies like Videotron made the point, saying that they can have dissuasive business practices to sort of make people think twice about downloading too much.
1857 On the other hand, from a commercial point of view, or an Industry Canada point of view, if a company advertises 10 megabytes per second, and only that, should you, as a customer, think that you are allowed to use that 10 megabytes per second?
1858 When you buy your phone line from Bell Canada, do they specify how many hours a week you are allowed to use it? No.
1859 You pick up the phone, you dial, and the number of times when you get fast, busy or no dial tones are very rare.
1860 Why is that? Because the telephone companies have adapted their phone systems over the years to changing user patterns.
1861 When dial-up internet came along, that caused big mayhem for the phone companies, because all of a sudden you had people staying two hours on the phone --
1862 THE CHAIRPERSON: Yes, but --
1863 MR. MEZEI: Let me just finish --
1864 THE CHAIRPERSON: I don't have all day for you, Mr. Mezei, you answered my question, okay? Thank you.
1865 MR. MEZEI: Okay.
1866 THE CHAIRPERSON: Mr. Roks, you say, "In acceptability of ITMP, any additional traffic measurement method should be under emergency circumstances, and only when network upgrades take place."
1867 That is sort of a nice edict to establish, but -- first of all, obviously, upgrades are very expensive, and secondly, surely the networks, if there is congestion, have to have an ability to preserve the integrity of the network, especially for those users for whom latency and jitters are an issue, as opposed to downloaders, for whom it may be an inconvenience, but it doesn't make the use of the internet impossible.
1868 MR. ROKS: What exactly is your question?
1869 THE CHAIRPERSON: I don't understand how you can just say categorically --
1870 MR. ROKS: Well, their problem is the network. The solution is to fix the network. To come up with other ways to avoid fixing -- upgrading the network aren't solutions, and they shouldn't be allowed unless there is clear --
1871 THE CHAIRPERSON: But, surely, upgrading costs money.
1872 MR. ROKS: Yes, it does.
1873 THE CHAIRPERSON: And if that money is not available, or is not part of their business plan, you are now saying that we should say, regardless, "That is your priority, that is where you have to put your money"?
1874 MR. ROKS: Then maybe they should stop signing up new customers, or let go some of their customers, because the network capacity that they have sold isn't sufficient to meet the demands and what they have sold the customers.
1875 If they can't upgrade their network because it costs too much money, then they should let some of their customers go, or allow someone else to build a network that could provide that service for them.
1876 THE CHAIRPERSON: Or tell the customer, "On my network you can only get capacity up to X, and if there is usage beyond that there will be throttling."
1877 MR. ROKS: I would think that, if there was competition in the marketplace, that wouldn't be something where a customer would stay. They would walk across the street to a competitor.
1878 Unfortunately, we don't have a competitor. We do not have one-to-one competition in this country.
1879 You have competing technologies that are being called competitive. DSL and cable don't compete. They both do broadband, but they are different technologies. For me to switch from one to the other, I have to replace my modem, I have to deal with different issues.
1880 The same with switching from, say, Bell's phone to Rogers' phone. One is using GSM and one is using another. That is not a valid competitive choice. I have to get a new phone. I am locked, based on the way competition is defined in this country.
1881 My point is, if they have a problem with the cost, let somebody else build them, or stop signing up new subscribers if your network is congested. You have hit your cap.
1882 If I have 20 oranges, I don't sell 40, I sell my 20 oranges.
1883 They've got 20 oranges; they are trying to sell 40.
1884 THE CHAIRPERSON: Len, over to you.
1885 COMMISSIONER KATZ: Thank you, Mr. Chairman.
1886 I am going to follow up on that frame of thought from a slightly different perspective; that is, one of your recommendations, Mr. Roks, is full disclosure, the transparency of all network management policies and practices.
1887 I think that there might be cases where consumers and users aren't fully aware of what they are paying for, and I think you made that point here a couple of times.
1888 Could you expand upon the concerns you have, and I will try to jump in and --
1889 MR. ROKS: I am not sure that I fully understand your question.
1890 COMMISSIONER KATZ: Your first recommendation: "Above all else, transparency of all network management policies and practices..."
1891 What you are talking about there is purely the issue of what is being throttled, and divulging that, that is one thing, but I think you have more than that in mind.
1892 MR. ROKS: And there is an exception on acceptable protocols, being security --
1893 I listed out three things that are acceptable, which are security, spam, and any kind of malicious behaviour on the network. So I am not saying that --
1894 There has to be some sort of management for malicious activity. I am not saying rule out everything, but anything that deals with the content itself or applications should not be allowed.
1895 COMMISSIONER KATZ: When you refer to WC3 standards for guidelines, what are you referring to there?
1896 MR. ROKS: WC3 is a body that establishes industry standard codecs, kind of like an ISO group, for example.
1897 They have a number of protocols, formats, that they accept as being standards.
1898 What that would do is, it would promote new protocols to go and get legitimized through organizations and groups.
1899 If that were the case, I am sure that the people from BitTorrent, which is an actual profit-based corporation incorporated in the United States, would make efforts to do that.
1900 COMMISSIONER KATZ: Your next recommendation speaks to the establishment of an internet ombudsman.
1901 Are you aware that there currently is a Commissioner of Communications Complaints service in Canada?
1902 MR. ROKS: I recently became aware of that, yes. I am just not sure what their jurisdiction is, and whether they have the resources and staff to be able to handle the complaints and actually follow up effectively.
1903 COMMISSIONER KATZ: Their role is to handle complaints for non-regulated services. So where there is a competitive market out there, if a consumer feels that he is either paying for something and not getting it, or there is an error in the way he is being treated, that is where they can turn to, and they are equipped to deal with it.
1904 And staffing, and everything else, obviously, works on a recovery basis. They are fully funded by the parties that they adjudicate on behalf of.
1905 So to the extent that a complaint --
1906 MR. ROKS: So it's industry managed.
1907 COMMISSIONER KATZ: I'm sorry?
1908 MR. ROKS: It's self-regulated.
1909 COMMISSIONER KATZ: That's right.
1910 The industry is paying for the infrastructure that is there to handle these complaints.
1911 MR. ROKS: But doesn't that create a bit of a conflict of interest?
1912 COMMISSIONER KATZ: No, because it's totally independent.
1913 What happens is, if you have a complaint and you go to Bell Canada and they can't solve the problem, you can then go to this commissioner, who will investigate it, and at the end of the year the total cost of their role is distributed amongst all of the players they regulate.
1914 It's not a one-to-one relationship, so there is no conflict.
1915 MR. ROKS: What you are saying is, if our people have complaints in regards to internet issues, they should contact the ombudsman.
1916 COMMISSIONER KATZ: That is the vehicle to go to, yes.
1917 MR. ROKS: That's good to know and have that clarified, because I wasn't aware of it before. Only recently, when I ran this by somebody else, did they mention that there is one and there has been some discussion about expanding the jurisdiction or the role of that individual.
1918 COMMISSIONER KATZ: One of the issues that comes up frequently in my discussions with folks day-to-day is the fact that the service providers are promoting their services with speeds and bandwidth up to certain capacity levels, or whatever, and then these things end up getting throttled or adjusted, whatever the case may be.
1919 Is there an issue from either one of your perspectives in how the services are being marketed today in terms of the disclosure?
1920 It could be zero, or it could be up to whatever they are proposing, as well.
1921 Is there perhaps a misrepresentation there? As opposed to promoting "up to" capacity levels, you promote minimum levels of capacity and speed?
1922 MR. MEZEI: In my presentation -- first, because of the time, I speak too much -- yes, I see a big problem with this.
1923 They have raised speeds to impress people, but they are not able to actually provide --
1924 Bell now has a service that is, I think, 16 megabits this week, and it might be 17 or 18 next week -- they keep changing it -- yet they were in front of you last year saying, "We don't have enough capacity, there is congestion."
1925 Last year, when they did the CAIP thing, their speeds were at 7 megabits per second, and now they have raised it, but they haven't really upgraded their network since then, because they are still complaining of congestion.
1926 They raised their speeds, and they can't provide that throughput. They really should --
1927 And I think that the Commission should put its foot down and tell the telcos, "If you want to throttle, you have to publish, in as big a font as your speed, the throttle speed."
1928 Bell Canada chose 30 kilobytes per second, which is 240 kilobits per second. That is not even one-tenth of what is advertised.
1929 If they were forced to advertise that, you might find that their throttle speed might go up, all of a sudden, magically, and their congestion problem might disappear.
1930 Really, has the Commission heard from many individuals complaining about congestion problems?
1931 You have heard from Bell and the big telcos. You have heard from what I call the legacy telcos complaining about it. You haven't heard from the small ISPs. They have managed their networks.
1932 The raising of the speeds is, in my opinion -- and I will be blunt -- false advertising, because they can't deliver.
1933 And from the Commission's point of view, I am sure you can legally find a way to force them to divulge something real. You have your experts.
1934 But, in my opinion, that is a strong point that needs to be made. That will give them incentive both to upgrade their networks and divulge what they are actually doing in terms of management.
1935 COMMISSIONER KATZ: From your perspective, what would be the reason, then, that they are continuing with their throttling, if, in fact, there is no concern by consumers?
1936 MR. MEZEI: My concern with this whole thing is not so much the congestion, but the fact that they picked on one application.
1937 If you look at the companies who have complained about P2P, they have a vested interest to not see this work, because they own cable television, they own satellite...
1938 Bell wants its IPTV service, which would be killed if people downloaded from the internet and did not pay the IPTV.
1939 This may be just propaganda from me or just conjecture, but it happens that it's just those companies complaining about P2P and nobody else. You don't see the small ISPs -- the small ISPs are more than happy to take on customers who want to do P2P. They can manage a network. They can buy the capacity to supply. They are not complaining, it's just the big guys who do.
1940 And the fact that they are focusing only on P2P is what worries me.
1941 COMMISSIONER KATZ: Mr. Roks, you mentioned earlier that the flavour today is P2P, it used to be Kazaa and whatever else, as well Napster or whatever --
1942 MR. ROKS: Actually, Kazaa was here appearing.
1943 Peer-to-peer, itself, is not a technology, it's a name given to a bunch of software that uses network-to-network connections, which is really the internet. It's all about point-to-point connections that then relate to another point, that relate to another point.
1944 So peer-to-peer is, in effect, what the internet is, and to say that there are peer-to-peer applications -- I mean, peer-to-peer is used in so many different ways. It is used to transport video. It is used to crunch data.
1945 Actually, there is one that uses peer-to-peer data to make screen savers and very high resolution graphics that render on 100,000 machines around the world, while your machine is sleeping in screen-saver mode.
1946 There are lots of different ways that you can use it, and it doesn't necessarily cause the problems on the network that the ISPs are saying it does.
1947 COMMISSIONER KATZ: Let me turn it around then. If peer-to-peer is the internet, then what is being discriminated?
1948 Everything is being discriminated; therefore, nothing is being discriminated, by definition.
1949 MR. ROKS: We don't know, because I think -- Bell's response was, "We are throttling all high bandwidth applications -- hub bandwidth usage applications," at one point. At another point they said that they were throttling peer-to-peer traffic at certain times of the day.
1950 It has never been very clear what exactly they are doing.
1951 Right now, as far as I see it, it is being selective, because I know that a lot of my other bits are getting through that use peer-to-peer technology such as Flash, when I am watching a video.
1952 I know that those are getting through at a decent speed, so they are being selective.
1953 So it's not about peer-to-peer, it's about certain peer-to-peer applications that, in my personal opinion, potentially introduce risk to their existing business models.
1954 COMMISSIONER KATZ: We are back to disclosure again. If they disclosed what they were doing, then there would be more information available to the public, assuming that it was --
1955 MR. ROKS: I don't think that they should be able to do it at all, but if they are going to do it, then it should be disclosed.
1956 And that's not up to me. If it was my decision, then, no, they wouldn't be allowed to do it. But if your decision is to allow them to continue those practices, then I think you have to ask that question.
1957 COMMISSIONER KATZ: Those are my questions, Mr. Chairman.
1958 THE CHAIRPERSON: Candice...?
1959 COMMISSIONER MOLNAR: Thank you.
1960 I just want to ensure that I understand -- you spoke a little bit about what we would call the consumption model, where consumers have to pay for bandwidth over a certain threshold, and perhaps that is one way of being able to afford the new consumption that is out there and is driving some of this congestion on the network.
1961 As I understood, you are not opposed to that type of model.
1962 MR. MEZEI: It depends on how it is implemented.
1963 Let's face it, the internet is a product -- it's a utility. Gordon Brown has basically admitted that it has to be a utility, and from a utility point of view, I think it is probably fair that people pay for what they use.
1964 Whether you pay by the byte, which is what Bell wants to do with its current tariff stuff, or whether you have different prices depending on what speed you are getting...
1965 Basically, right now, if you have a lower speed -- modem speed that they give you, let's say 1 megabit per second, first of all, you are not going to download large movies, because it's too slow to do it, and you are not going to cause as much congestion on the network because you can never burst that high speed because your modem limits you.
1966 So there is a way to limit without being usage-based, basically.
1967 I think that you would find that a lot of people, a lot of citizens, are afraid of price plans where they don't know what the amount is going to be at the end of the month, and by having a speed limit, or a speed-based plan, like you use, you give the consumer the opportunity to have a fixed cost per month without abusing the network.
1968 When you look at what is happening with the internet now, as you increase speeds, you enable new applications, new innovations, downloading HD movies, and God knows what else is coming 15 years down the road, but you still have your grandmother, who probably just wants to download her e-mails once a day. She shouldn't have to pay for the capacity, or the ability to download virtual 3D worlds, or whatever, you know, that will take 50 gigabytes.
1969 The ISPs can produce price plans for different types of uses without being discriminatory in the way they manage their network.
1970 Right now, I think that is where we are getting to, where the networks want to advertise nice prices, and not come down to say, "Pay for what you use," and come down with, basically, open -- clear plans saying, "If you want to use this, you buy more. You pay more, you get more. But what you pay for, you get all of it."
1971 COMMISSIONER MOLNAR: Yes, I understand your point on disclosure. That was very clear.
1972 We heard that, basically, metered pricing, a price plan consumption model, may work well for what was defined as the steady state, the ongoing usage of the network, but it is not adequate on its own to address peak or unusual congestion in the network, so other types of traffic management may also be necessary.
1973 You have been clear that you don't support application-based traffic management, something that targets a specific application.
1974 MR. MEZEI: Correct.
1975 COMMISSIONER MOLNAR: Are there any technology solutions that you view to be more reasonable?
1976 MR. MEZEI: Yes, because those fancy boxes that Bell installed, or that the other telcos -- the DPI equipment, they are fancy computers basically, slash routers, with two wires, one coming in and one going out. They can be programmed to look at how much data you are using live. They could be programmed to do that, and I suspect that they do.
1977 Right now they have packages that they want to sell that do the DPI stuff, that look into the packet to pick out certain products and throttle them.
1978 The industry is large enough that if there is demand to limit the speed, it gets done very quickly, and the VIF example, I think, is great.
1979 I am really glad that I heard on the internet yesterday -- it wasn't throttled -- I heard the speeches yesterday, because I picked up on it, and I figured, well, I have added that slide in there. They have done this for 12 years with standard equipment, nothing special.
1980 And the routers have long been able to do that.
1981 This is a decision that the ISP can make, "I am going to have a price plan," and they can manage it.
1982 Now, to get back to usage-based billing preventing peaks, you can't be perfect in the world. To say that we don't want any congestion, I think, would be ludicrous.
1983 The internet was built to handle congestion. You have your 100 megabit Ethernet at your home from your PC to your modem. Your modem is at 5 megabits. There is congestion right there, and it learns to cope with it.
1984 You are sending your data, initially, at 100 megabits per second, and your computer quickly realizes, "Whoops, that's too quick," and it slows down.
1985 It works that way. It has worked. So you have a few times where you have peaks, but I haven't heard people complain about congestion. People don't actually notice this.
1986 The networks are complaining about it, but how many complaints has the CRTC received about congestion?
1987 One last thing, on the actual internet, where you have transit providers, their business is to provide bandwidth, and they like to sell bandwidth because they make more money, and when they sell bandwidth they upgrade their network, and they have no problems.
1988 Let me tell you this, a network called Cogent started off as sort of a run-of-the-mill PSI network, which was nearly bankrupt. It had a very bad reputation. It had congestion and it started to lose customers because there was competition. They went to other transit providers. Cogent fixed its image, fixed its network, upgraded it, and now they are reliable.
1989 COMMISSIONER MOLNAR: Just to interject, you do not support any technology-based solutions for traffic management, it's consumption based, address consumers' use, no technology, or application-agnostic approaches from a technology basis, simply allow congestion if that is the consequence in peaks.
1990 MR. MEZEI: There needs to be a balance. I don't support application management, so the solution that is permitted has to be application-agnostic. It has to be content-agnostic.
1991 There can be a certain amount of congestion that is allowed, in the same way that the phone system -- I suspect that a couple of times a year you may pick up the phone and not get your dial tone for a few seconds. That is sort of expected.
1992 What is reasonable? I am not in a position to tell you what percentage is reasonable, but a certain level should be reasonable. It doesn't need to be an all -- you know, solves all problems.
1993 But, really, from a country point of view, from the Telecom Act point of view, you can't look at packet contents, and you can't decide: I am going to throttle this application, because I have looked at the content and I don't like it.
1994 Basically, that is what Bell is doing.
1995 THE CHAIRPERSON: You are coming quite close to the revisiting of CAIP, and I don't want that, that's not here.
1996 MR. MEZEI: Okay.
1997 THE CHAIRPERSON: Suzanne, you have more questions?
1998 CONSEILLÈRE LAMARRE : Monsieur Mezei, je veux revenir sur... vous avez parlé de VIF Internet là. Si je comprends bien, VIF Internet, qui est à Montréal, c'est un fournisseur de services Internet qui aurait mis en place un système qui permet de limiter la consommation de ses clients; c'est ça?
1999 M. MEZEI : Bien, écoutez, limiter la consommation des clients, c'est des systèmes qui existent depuis des années, depuis toujours essentiellement...
2000 CONSEILLÈRE LAMARRE : Bien là, je veux parler seulement de VIF Internet.
2001 M. MEZEI : De VIF Internet, O.K. Moi, quand je les ai appelés hier, ils m'ont dit, ça fait 12 ans qu'ils ont ça.
2002 CONSEILLÈRE LAMARRE : O.K. Donc, vous ne savez pas si c'est un système qu'ils manufacturent et qu'ils peuvent vendre à d'autres fournisseurs Internet?
2003 M. MEZEI : Non, c'est des routers off the shelf.
2004 CONSEILLÈRE LAMARRE : O.K. C'est une combinaison d'équipements qu'ils utilisent?
2005 M. MEZEI : C'est des équipements qui sont disponibles depuis très longtemps. C'est juste quand tu le configures, tu dis, bien, tel range d'adresses IP...
2006 CONSEILLÈRE LAMARRE : O.K.
2007 M. MEZEI : ...ont une vitesse moins basse, ils ont une priorité plus basse, ça fait que quand il y a de la congestion, c'est eux qui sont touchés. Puis essentiellement, les clients sont * punis + parce qu'ils ont déjà trop consommé, mais ça leur permet de gérer leur réseau comme ça. Et c'est sur leur site Web, c'est documenté et c'est ouvert.
2008 CONSEILLÈRE LAMARRE : O.K. Merci pour la clarification.
2009 THE CHAIRPERSON: Mr. Roks, a final question. You said Bell refuses to peer. What did you mean by that? I mean surely peer-to-peer applications work over the Bell network.
2010 MR. ROKS: Okay. Well, peering and Internet exchanges, even though it uses a similar word, "peer," is not the same as peer-to-peer. It uses the word "peer" because a lot of what the Internet does is about peering and connecting.
2011 So Bell -- the way it works is there are things in every city called a local loop. It is a non-profit organization. In Toronto it is called TorX, in Quebec it is called QueX. There is one in Vancouver. There is one in Edmonton.
2012 What happens is all the ISPs connect together to loop their data together. So if I am on Rogers and someone else is on Videotron, rather than it going out to the Internet, which costs both ISPs money, it goes directly over the local loop and delivers at zero cost.
2013 THE CHAIRPERSON: Right.
2014 MR. ROKS: So all the money you are saving in that, you actually can put into network upgrades, especially when you are billing people on bandwidth --
2015 THE CHAIRPERSON: Sure.
2016 MR. ROKS: -- or make it clear what is on-LAN and what is off-LAN traffic.
2017 So Bell refuses to peer with anybody. With Rogers you have to be doing 5-megabit sustains at any given time.
2018 THE CHAIRPERSON: Peer means sharing this local loop?
2019 MR. ROKS: Meaning sharing on the local loop.
2020 THE CHAIRPERSON: I see.
2021 MR. ROKS: There are many examples out there. TorX is probably the best example in Canada. Today, as I said, they are doing 15 gigabits per second over that network, which is Akamai, Google, CBC is connected to that. CBC is running Torrent off that as well and that doesn't cost them any money to distribute in Canada and it is being delivered in a more efficient manner.
2022 THE CHAIRPERSON: Okay. Thank you very much for that clarification.
2023 MR. MEZEI: Can I make just a quick comment on that? In terms of the peering, just to give you an example, as he said, from one small ISP to another, if they peer, their connection is direct.
2024 THE CHAIRPERSON: Yes.
2025 MR. MEZEI: But for a small ISP to connect to its bank, who is hosted by Bell, the connection has to go to Chicago, to New York, and at New York they connect to Bell and back to Canada. It is very inefficient because Bell doesn't want to connect directly within Canada. But that is outside, I guess, the purview of this hearing.
2026 MR. ROKS: I will give you one more example on top of that. If I am playing a game from my Rogers connections with someone on a Bell connection, the latency is very high because Rogers now has to send their data over the Internet as opposed to directly to them. So overall, the experience becomes poor because Bell won't peer.
2027 THE CHAIRPERSON: And you only mention this because you are basically saying there is a way that Bell could be more efficient without expending money but they refuse to do it for some reason?
2028 MR. ROKS: Yes. They could participate in the overall management of the Canadian network as opposed to just their own.
2029 THE CHAIRPERSON: Okay. Thanks very much for your presentation.
2030 MR. ROKS: Thank you.
2031 LE PRÉSIDENT : Madame la Secrétaire, à quelle heure est-ce qu'on commence demain?
2032 THE SECRETARY: We will reconvene tomorrow morning at 9:00 a.m.
2033 THE CHAIRPERSON: Okay. Thank you.
--- L'audience est ajournée à 1428, pour reprendre le mercredi 8 juillet 2009 à 0900
Johanne Morin Jean Desaulniers
Sue Villeneuve Beverley Dillabough
____________________ ____________________Monique Mahoney Madeleine Matte
- Date de modification :