p2pnet news view P2P | Freedom:- “In accordance with the Commission staff letter of 19 June 2008 and Bell Canada’s (or the Company’s) letter of 9 July 2008, the Company is hereby filing its Answer to the CAIP (Canadian Association of Internet Providers)  Application dated 3 April 2008.”

That’s Bell Canada’s Mirko Mr 5% Bibic to Canadian Radio-Television and Telecommunications Commission secretary general Robert A. Morin.

Bell admits it’s, “also in receipt” of 25 comments from such as Google and Per Vices, the company which also provided a way for people to get around Bell’s bandwidth throttling “traffic management” actions. The complete list is in the document below.

Dear Mr Morin, it says >>>

Subject: Application requesting certain orders directing Bell Canada to cease and desist from “throttling” its wholesale ADSL Access Services

In accordance with the Commission staff letter of 19 June 2008 and Bell Canada’s (or the Company’s) letter of 9 July 2008, the Company is hereby filing its Answer to the CAIP Application dated 3 April 2008.

Sent in yesterday, it’s a day late —- Bell asked for, and was granted, an extension —- and considerably more than the dollar short.

With a little luck, we’ll be able to give you CAIP’s response on Monday or Tuesday.

But here it is in full, including an interesting dissertation on deep packet inspection (DPI).

I believed it was important to have this document available as soon as possible and I apologise in advance for any mistakes I made in converting by hand it to html, for any other mistakes which may be down to me, and for the fact I’ve left footnote numbers in. [Please note, Figure 10 (the number of congested links per month for the period of March 2007 to May 2008 for each location in the network ) is missing in our copy. But figure 11 expresses the same data as a numerical percentage rather than absolute numbers. Figure 13 is also missing.]

Click here  for the full document (as I received it), including footnotes, and in the meanwhile >>>

BEFORE THE CANADIAN RADIO-TELEVISION AND TELECOMMUNICATIONS COMMISSION IN THE MATTER OF AN APPLICATION BY CANADIAN ASSOCIATION OF INTERNET PROVIDERS PURSUANT TO PART VII OF THE CRTC TELECOMMUNICATIONS RULES OF PROCEDURE AND SECTIONS 7, 24, 25, 27, 32, 36 AND 62 OF THE TELECOMMUNICATIONS ACT REQUESTING CERTAIN ORDERS DIRECTING BELL CANADA TO CEASE AND DESIST FROM “THROTTLING” ITS WHOLESALE ADSL ACCESS SERVICES

ANSWER BELL CANADA
11 JULY 2008

1.0 Executive summary

ES1 In accordance with the Commission staff letter of 19 June 2008, the following constitutes the Answer of Bell Canada (or the Company) to the CAIP Application dated 3 April 2008.

ES2 Bell Canada’s GAS customers are an important and valuable segment of Bell Canada’s wholesale business and the Company endeavours to supply them with the best possible network so they can provide internet services to their customers. In order to do so, the Company must, like any responsible network owner, manage its network. During the course of this proceeding, a number of misrepresentations and unsubstantiated allegations have arisen (which, it is important to note, do not necessarily reflect the opinions of the majority of Bell’s GAS customers). The Commission’s assessment must be based on facts, not unsubstantiated allegations or conjecture. The Company’s Answer is based on facts, data pulled from extensive studies on the growth of Internet demand, network capacity and congestion, as well as the Company’s more than a decade of experience in managing an Internet network and over a century’s experience in managing a telecommunications network.

Congestion

ES3 Internet traffic has grown exponentially and is expected to continue. For example, Cisco predicts that from 2007?2012 Internet traffic will grow six fold worldwide. Such growth does, and will continue to, put strain on network capacity. The Company has provided the number of congested links per month for the period of March 2007 to May 2008 for each location in its network (i.e., central office DSLAM, Aggregation network, BAS and Backbone network) as a numerical percentage. The total percentage of all four types of congested network links during a given month in the period in question has varied between 2.6% and 5.2%. While these numbers may seem low to the average lay person, they are significant to network traffic engineers such that it is important to consider the number of congested links in the proper context.

ES4 Since end-users’ high-speed traffic travels across multiple links in the Company’s network, if a single link along the way is congested, an end-user will experience negative impacts of congestion such as a slow connection or inability to connect. Once congestion has been observed, this means that latency has already been occurring in the network thereby producing negative impacts on customers. This phenomenon is analogous to a road system. When some of the major arterial roads are congested (analogous to Backbone links), cars travelling from the suburbs to downtown are impacted by traffic regardless of the state of congestion on the roads in the suburbs (similar to DSLAM and Aggregation links). Just like a single traffic roadblock can hinder drivers going to multiple destinations that pass through the road that is blocked, a very small amount of congested links can seriously affect a large number of high-speed end-users’ traffic. When traffic levels increase, the potential for drivers to be impacted by more than one roadblock also increases. The impact of that congestion is clearly perceptible to the end-user, though to varying degrees. The effects are even more obvious in the case of time sensitive applications such as voice communications, where the content can become severely degraded to the point of being unrecognizable.

ES5 Depending on the type of link, the potential number of impacted customers will vary. The Company had projected that the number of customers that could be affected by congested links could reach as high as 790,000 by the end of the first quarter of 2009 were Bell Canada to fail to apply its network management solution.

The Company’s Solution

ES6 In order to address network congestion, Bell Canada has essentially adopted a three-pronged approach to managing capacity on its network:

i)Investing in capacity through managed capital spending

a.Since 2001, the Company has invested over $3 billion in capital investments on its high-speed Internet service. The Company plans to spend close to half a billion dollars in 2008 to expand and upgrade the infrastructure of its DSL network.

b.In 2004, the Company started building its next generation DSL network using Fibre to the Node technology (FTTN) and Optical Ethernet (OE) in its aggregation network. When relief is required to address congested links on the ADSL ATM network, the Company is required to invest in legacy technology to support the ATM network. While ATM is both expensive and limited in its capacity when compared to OE, nonetheless, the Company continues to make those investments on behalf of its customers. The Company is increasingly investing in its Ethernet aggregation network as well as its FTTN network, but it will take time until it is expanded and available to the majority of Internet end-users. During this proceeding parties have continuously oversimplified the work required to address network congestion. In paragraphs 98 to 101, the Company notes several inconsistencies and faulty assumptions in an article1 presented by CISP to support its claims. In the article, Mr. David Burnstein, a DSL industry analyst, ignored some very important costs. Contrary to his assertions, simply upgrading DSLAM ports to GigE is not all that is required. This may explain how the author formulated the incorrect conclusion that Bell Canada’s congestion is minimal and easily solved.

ii)Moving toward usage-based pricing and new business models in a manner that benefits end-users and that takes into account the realities of the highly competitive marketplace

a.The Company initiated the transition to usage-based billing in December 2006 when it ceased offering its ‘unlimited traffic’ plan to any new subscribers. However, the transition to usage-based pricing cannot be the sole solution to network congestion, nor is it an instant “fix” for three principal reasons:

i.Transitioning from “unlimited traffic” pricing to “usage based” pricing will take time; from a purely contractual basis, all customers would have to be adequately notified of such a significant change; from a systems viewpoint, the Company would have to design, build, implement and seamlessly transfer customers from unlimited billing to billing that indicates usage based prices.

ii.Carriers will need to carefully experiment with different pricing plans in the marketplace in order to maintain customer satisfaction and thus, maintain its customer base. This will involve a high degree of trial and error which necessitates market trials rather than mass migrations of its entire customer base.

iii.And for other competitive reasons that have been submitted in confidence to the Commission at paragraph 106 of the Company’s main submission.

b.Even once the transition to usage-based pricing is achieved, usage pricing alone will not suffice in reducing network congestion. For example, as consumption continues to increase, necessitating further network investments, network management actions will be required as pricing plans are adjusted to respond to these new demands on the network. There will also likely remain a need to manage “bursts” of traffic which usually, but not always, occur during “peak” periods.

iii)Managing bandwidth

a.For the above reasons and those set out in detail in this submission, “managing bandwidth” must remain an essential component of Bell Canada’s Internet traffic management solution. The solution includes the following:

i.terminating or managing the service of users who consistently breach the Company’s Acceptable Use Policy; and

ii.better balancing Internet traffic on its network during peak periods by using DPI to redistribute P2P file sharing traffic to off-peak periods. The Company does not block any form of P2P file sharing applications nor does it shape non P2P file sharing applications.

ES7 The Company has not arbitrarily chosen to shape P2P file sharing applications. There are two reasons why P2P file sharing applications are subject to traffic management measures. First, P2P file sharing is not a time-sensitive application. As noted below, the Company does not traffic shape time sensitive P2P applications such as Skype and Joost. Second, P2P causes unique problems for network providers. The attraction of the P2P protocol from a content and application provider perspective (i.e., the distributive nature of the service which reduces costly storage and centralized upload capacity) is the very same attribute that creates significant problems for the network provider. Due to the nature of its architecture there are three ways that P2P file sharing applications use a disproportionate amount of bandwidth compared to other types of traffic:

i)P2P file sharing traffic is designed to open several sessions in an effort to transfer data as fast as possible, thus overwhelming other forms of traffic.

ii)Because of the possibility of queuing file requests, P2P file sharing can sustain a continuous maximum network traffic load, 24 hours a day, 7 days a week and 365 days a year, as long as there are queued requests.

iii)Some P2P file sharing applications look for the fastest node available, and thus any increase in capacity to one network node will attract increased P2P file sharing upload requests from other P2P file sharing applications resident on other networks. As described by Rogers’ Chief Strategist at the latest Telecom Summit, Rogers’ tests have indicated that an increase of capacity at a node could be eaten up by P2P file sharing applications within 24 hours. Indeed, the Company’s own testing shows that in some cases the increase in capacity could be eaten up in as little as 30 minutes.

ES8 As P2P file sharing applications are designed to use all of the bandwidth that is available, additional capacity cannot, on its own, resolve this issue. Network management is currently the only realistic response and has been applied, on an objective and rational basis, to both the Company’s retail and wholesale GAS services. Because P2P file sharing applications are non time-sensitive, these can be “slowed down” during peak hours of traffic without interrupting service. Additionally, the bandwidth that is freed up allows real-time applications to make use of this bandwidth. The Company does not shape other types of traffic, including that of non P2P file sharing applications. P2P file sharing applications are the only signatures for which traffic shaping policies apply – these applications are easily distinguished by the DPI from other applications such as Skype or Joost. Accordingly, Skype and Joost traffic are not being shaped.

Legal issues raised in CAIP’s Application

Network management activities permitted by tariffs

ES9 The GAS Tariff contains the terms and conditions pertaining to the GAS service. Contrary to CAIP’s allegations, the Company has not acted outside of the bounds of the Company’s Tariffs. Rather, the Company’s traffic management measures are performed in a manner that is consistent with its tariffs, its contractual obligations and the Commission’s determinations in this regard. The Commission has long recognized that the Company’s Terms of Service, found in General Tariff – 6716, Part 1, Item 10 (Terms of Service) and applicable to the Company’s GAS service:

“…allow them to suspend or terminate service where a customer uses or permits others to use services so as to prevent fair and proportionate use by others.”2

The Commission has accordingly stated that it is not necessary to include explicit “network congestion” provisions in specific telephone company tariffs as such provisions are already accounted for by virtue of their respective Terms of Service.3

No unjust discrimination

ES10 The Company notes CAIP’s allegation that the Company’s traffic management measures are contrary to section 27 and in particular, subsections 27(2), (3) and 4 of the Telecommunications Act (the Act). Allegations similar to those of CAIP in the present proceeding were made by Cybersurf Corp. in a proceeding against the cable companies. Cybersurf, a reseller of Shaw’s High-Speed Internet Service (HSIS) asked the Commission to intervene on the basis that Shaw was conferring an undue preference onto itself by making Quality of Service Enhancement service (QSE) available to its own retail customers without reselling it to Cybersurf. The Commission denied Cybersurf’s petition in that case:

The Commission finds that based on the evidence provided in this proceeding, no traffic is given priority over other traffic, and in particular, there is no evidence that Shaw’s QSE gives its traffic preference over Cybersurf’s or any other competitor’s traffic on the HSIS network. The Commission therefore cannot find that Shaw is conferring onto itself an undue preference by offering its HSIS customers the QSE while not offering the same to Cybersurf for resale to its end-customers.4 [Emphasis added]

ES11 The Commission also addressed nearly identical issues to those in the present instance in rendering its determinations in Telecom Decision CRTC 2006-77:5

248. The Commission considers it appropriate that each cable carrier be provided the ability to manage the potential negative outcome of high-consuming bandwidth end-users in a manner that does not degrade the Q of S to all end-users, whether it is the cable carrier’s end-user or the competitor’s end-user. The Commission considers, however, that regardless of the approach adopted by the cable carriers to address this problem, such an approach must provide equivalent treatment with respect to excessive usage to both its own retail Internet access service end-users and TPIA end-users.

…

249. With respect to the different treatment by cable carrier of end-users for excessive usage, the Commission considers that, subject to the requirement for equivalent treatment specified above, each cable carrier is permitted to implement its own approach. [Emphasis added]

ES12 GAS service is not offered over a separate and distinct network. The GAS service offered to wholesale ISPs and the retail Sympatico Internet service offered by the Company share and have always shared the same access network and therefore will experience the same benefits and the same problems. As the Company is applying the same treatment to traffic of its own retail Internet access service end-users and GAS end-users, the Commission cannot find that the Company is conferring unto itself a preference, let alone an undue preference.

ES13 The Company further notes that the implementation of the Company’s traffic management solution has not resulted in a substantial lessening of competition.

i)The high speed Internet access market is highly competitive.

ii)The Company’s number of total GAS end-users has continued to grow each month before and after traffic shaping was implemented.

ES14 Finally, there is no ulterior motive attached to the Company’s management of network congestion; it was not launched in order to i) launch usage-based billing, ii) facilitate the launching of an IPTV service or iii) give an undue preference to the newly launched Bell Video store nor for any other purpose unrelated to the management of network congestion.

ES15 Despite the many claims by CAIP and other parties, the evidence clearly demonstrates that there is no ulterior motive or any basis for a claim of unjust discrimination. The Companies note the particular irony in Google accusing Bell Canada of playing the ‘gatekeeper role’ by traffic shaping P2P6 and thereby impeding competition. If there is, indeed, any gatekeeping activity on the Internet, which is questionable, the gatekeeping is being performed by the Internet search engines, which are typically the users’ ‘window’ to the near-infinite content available worldwide. And Google is widely acknowledged as being the dominant Internet search engine. In any event, it would be counter productive to impose regulatory rules to prevent traffic management techniques that are designed to improve the overall end user Internet access experience and to ensure that congestion does not harm innovation in the future.

Notification of network changes is not an issue

ES16 The Company submits that the Commission’s notification of network change requirements (the Notification Requirements) do not apply to the current situation for two reasons. First, the Notification Requirements do not apply to GAS service (as it is not a bottleneck service); second, traffic management is not a notifiable change because it does not require ISPs to make adjustments in their network.

Bell respects the privacy of its customers

ES17 The Company’s use of DPI technology as part of its traffic management practices is such that the actual contents of the communication exchange are not examined. Rather, only the protocol headers are examined and the DPI equipment does not retain the information reviewed in the packet headers.

ES18 While the Company has explained in detail, in section 10.4 of its submission, what its DPI devices are capable of doing; it is helpful to summarize what they do not do. By design, the DPI devices deployed in Bell Canada’s network do not:

- use any personal identification information of an individual user;

- store or log any personally identifiable information;

- have specific knowledge of a user’s real identity;

- have knowledge of a user’s content;

- have knowledge of a user’s URL browsing history;

- have knowledge of a user’s Internet search activity;

- have knowledge of a user’s email topics or content;

- store content accessed by a user;

- cache any content, including user-specific content, whatsoever;

- capture and playback any communications exchange; or

- install or require any specific software on user machines.

Bell is not in violation of section 36 of the Act

ES19 Bell’s DPI traffic shaping activity does not “control the content or influence the meaning or purpose of telecommunications” for two key reasons. First, slowing the delivery of content does not amount to “controlling” it. Second, Bell is not involved in any way with the editorial control of content being transmitted through P2P file sharing applications nor is it creating or preventing access to such content. The shaping technique is content and content provider agnostic. Bell cannot influence the meaning or purpose of the telecommunications because Bell has no knowledge of the content itself.

Unwarranted policymaking and/or regulatory measures could hurt innovation, impede competition and decrease the efficiency of Canadian telecommunications

ES20 The Company urges the Commission to exercise caution as it assesses the unsubstantiated allegations that have been made in support of requests for new and unneeded policies with regards to this tariff dispute. As remarked by the Commissioner of Competition at this year’s Telecom Summit:

The reason for intervention should not be so vague as “finding an appropriate balance.” Balance of what? Why cannot that balance be created ultimately by market forces? Often the pursuit of “balance” is little more than the protection of the status quo. Those who enjoy a market advantage will say that we have struck the right balance. They do not want more regulation, but neither do they want deregulation to open the market to new competitors. Is protecting the status quo the best way to increase innovation and prosperity? If you believe that, then state it, and let the debate begin. […]regulation should always be viewed, not as a first step, but as a last resort. [Emphasis added]

ES21 Several parties have claimed that traffic management will hinder innovation. The Commission should see these claims for what they really are: the use of policy rhetoric devoid of substance in order to promote vested business interests. Indeed, the opposite of these claims is clearly the case; necessity is the mother of invention. Without any regard to network capacity, P2P file sharing application designers could develop applications designed to use all of the bandwidth that is available. It is no coincidence that P4P is emerging as a possible solution with regards to P2P file sharing at a time when ISPs around the world are starting to implement traffic shaping measures. Indeed, P4P is still very much a nascent framework that has yet to be deployed for widespread use or as a viable solution for the Company’s use. The Company is confident, however, that competition and market forces will encourage, as it has already done, the development of new, better and fairer P2P file sharing applications.

ES22 Until that time, the Company has endeavoured to limit its traffic shaping to what it sees as a reasonable measure in order to address network congestion. It has implemented this solution prior to the network’s succumbing to “rampant congestion” as it seems that some parties would prefer that the Company do. It has opted, in an objective and rational manner, to direct its network management solution solely to P2P file sharing traffic and not other forms of traffic. Finally, it has opted to engage traffic shaping only during peak traffic hours and not 24 hours a day.

ES23 As a final note of caution, the Company notes that the Commission has asked the Company in the interrogatory stage a number of questions related to the state of congestion in its network. Further, in its Interim Answer and in this submission, the Company provided very detailed explanations of its Internet traffic management solution, a large amount of which Bell Canada has chosen to place on the public record. Its decision to offer these detailed explanations (mostly on the public record) is in response to the large interest in the media and the blogosphere, and to correct the miscomprehension about how Bell Canada is actually managing its network and the impact that it is having on its Internet end-users (retail and GAS customers). GAS customers generally, including the Applicants, have used and continue to use unsubstantiated allegations regarding the facts to then jump to inappropriate conclusions. The Company understands the Commission’s need to examine the detail of this congestion in order to satisfy itself that there is no unjust discrimination or anti-competitive motive underlying the Company’s actions. Outside of that one concern, the Company submits it is not for the regulator to second guess the Company’s engineering decisions. As the Commission has stated: “a carrier should be free to implement the standards and specifications of its choice within its network. To require otherwise could be costly and inefficient to some carriers, while possibly decreasing opportunities for product differentiation.”7 The Company has established that there can be no validity to any claims of unjust discrimination or anti-competitive conduct. The Company therefore submits that there is absolutely no need for Commission intervention in this case.

Conclusion

ES24 For all of the reasons above, the Company asks that the Commission deny the relief requested at paragraph 116 of CAIP’s Application.

2.0 Introduction

1.Bell Canada (Bell or the Company) is in receipt of an Application (the Application), dated 3 April 2008, made by the Canadian Association of Internet Providers (CAIP) pursuant to sections 7, 24, 25, 27, 32, 36 and 62 of the Telecommunications Act (the Act) and Part VII of the CRTC Telecommunications Rules of Procedure (the Rules) requesting that the Commission issue certain orders directing Bell Canada to cease and desist from “throttling” wholesale ADSL services and in particular, the wholesale service known as Gateway Access Service (GAS).

2.The Company is also in receipt of comments by:

Acanac Inc. (Acanac), 3 July 2008

Advancing Democracy & Media Sanity in Canada (ADMS), 3 June 2008

British Columbia Civil Liberties Association (BCCLA), 12 June 2008

Canadian Advanced Technology Alliance (CATA), 3 July 2008

Canadian Association of Voice Over IP Providers (CAVP), 12 June 2008

Canadian Internet Policy and Public Interest Clinic (CIPPIC), 21 May 2008 and 3 July 2008 (on behalf of the Campaign for Democratic Media (CDM))

Cisco Systems, Inc. (Cisco), 12 June 2008

Coalition of Internet Service Providers Inc. (CISP), 3 July 2008

Daniel Matan, 24 June 2008

Distributel Communications Limited, 3 July 2008

Google Inc. (Google), 3 July 2008

Information Technology Association of Canada (ITAC), 30 June 2008

Interactive Advertising Bureau of Canada, 3 July 2008

Kaboose Inc, 9 June 2008

Per Vices Corporation (Per Vices), 2 July 2008

Primus Telecommunications Canada Inc. (Primus), 15 April 2008 and 3 July 2008

Public Interest Advocacy Centre (PIAC), 3 July 2008

Rogers Communications Inc. (Rogers), 3 July 2008

Skype Communications s.a.r.l. (Skype), 11 June 2008

TCPub Média Inc. (TCPub), 9 June 2008

TELUS Communications Company (TELUS), 3 July 2008

Union des Consommateurs (UdC), 24 April 2008 and 12 June 2008

University of Western Ontario, 11 June 2008

Vaxination Informatique (Vaxination), 4 April 2008 and 3 July 2008

Wireless Nomad (Nomad), 22 April 2008

3.The Company hereby submits its answer in accordance with the procedures established by the Commission in its letter dated 19 June 2008. Pursuant to section 39 of the Act certain information in this response is being filed in confidence with the Commission. Absent a ruling from the Commission requiring disclosure, release of some of the information would be in contravention of Article 11 of Bell Canada’s Regulated General Terms of Service and Article 9 of Bell Canada’s Unregulated Terms of Service because it would disclose confidential information related to certain of Bell Canada’s competitors. Furthermore, release of the information would be in contravention of confidentiality requirements of applicable CSG agreements. In addition, release on the public record of certain information provided in this response would allow existing and potential competitors to formulate more effective business plans and marketing strategies, thereby prejudicing Bell Canada’s competitive position and causing specific direct harm to Bell Canada. An abridged version is provided for the public record.

4.Failure by the Company to address herein any issue or matter raised in any of the identified comments or those filed by other parties should not be construed as agreement to the extent that such agreement would be inconsistent with the interests of the Company.

3.0 Background

5.As part of its Application, CAIP also made a request for interim relief on an expedited basis requesting that the Commission issue an order directing Bell Canada to immediately cease and desist from traffic shaping GAS. Bell Canada filed an answer on 15 April 2008 (the “Company’s Interim Relief Answer” or the “Interim Answer”) solely addressing CAIP’s request for interim relief and factual discrepancies contained therein. CAIP filed its Reply to the Interim Answer (the Interim Reply) on 24 April 2008. The Commission denied the interim relief requested by CAIP on 14 May 2008 in Telecom Decision CRTC 2008-39, Canadian Association of Internet Providers’ request for interim relief regarding Bell Canada’s practice of “throttling” its wholesale ADSL access services. The Commission issued a process letter and interrogatories to CAIP and Bell Canada on 15 May 2008 with regards to the Application. On 29 May 2008, CAIP and Bell Canada filed their responses to the Commission’s interrogatories.

6.Certain parties, such as Google, PIAC and UdC, have also attempted to broaden the scope of this proceeding into a general “net neutrality” hearing. The Company agrees with Primus and TELUS that such issues are beyond the scope of this wholesale tariff dispute. However, while this dispute relates to wholesale GAS service, the Company will also discuss, in order to favour public debate, its Sympatico retail services due to the close relationship between the two and the fact that the Company’s network management practices are applied throughout the Company’s DSL network, including to its own retail services.

7.It must be noted that the Company, as a Canadian Carrier, is in control of its own network and is free to implement the standards and specifications of its choice within its network, as has been recognized in the past on several occasions by the Commission.8 Nevertheless, Bell Canada, in both the Interim Answer and in its 29 May 2008 interrogatory responses, provided very detailed explanations of its Internet traffic management solution in order to favour public debate and address a large amount of confusion and miscomprehension about how Bell Canada is actually managing its network and the impact that it is having on its Internet end-users (retail and GAS customers). Bell Canada’s GAS customers are an important and valuable segment of Bell Canada’s Wholesale business and the Company endeavours to supply them with the best possible network so they can provide Internet services to their customers. In order to do so, the Company must, like any responsible network owner, manage its network. Some GAS customers, including the Applicants (which, it is important to note, do not necessarily reflect the opinions of the majority of Bell’s GAS customers), have made and continue to make unsubstantiated allegations, and misinterpret the facts, to then jump to inappropriate conclusions. In assessing CAIP’s GAS tariff complaint, the Commission’s assessment must be based on facts, not unsubstantiated allegations or conjecture. Below the Company will set out the relevant facts to enable the Commission to reach a fully informed decision.

8.In paragraph 25 of its Application, CAIP claimed that that there was simply no congestion problem associated with the Bell Canada network, and thus traffic shaping is not required. As stated in its Interim Answer and interrogatory responses, Bell Canada rejects this allegation entirely. In this submission, the Company once again, and in even greater detail, describes the nature of the congestion issues and the steps it has taken to manage its network. In particular, the Company will provide details on the implementation of Deep Packet Inspection (DPI) technology and traffic shaping as part of its network management solution. The Company will also address third parties’ statements and correct several misconceptions regarding DPI. In order to address CAIP’s and other parties’ allegations, the Company has organized its answer as follows.

9.First, the Company will provide a description of GAS and how it is provisioned over the Company’s network. In describing GAS, the Company will address certain fallacies that have been promulgated through CAIP’s allegations.

10.In order for readers to fully understand the issues, the Company will follow with a description of the technologies involved including: a description of how traffic is routed on the Internet, how devices and applications on the Internet handle congestion followed by a description of peer-to-peer (P2P) technology.

11.Next, the Company will describe the growth in Internet demand, both globally and for Bell Canada, as well as the resulting increase in network congestion and how the Company is managing this increase in demand, including a description of the Company’s use of DPI for traffic shaping.

12.The Company will then address the legal issues that were raised in CAIP’s Application, and specifically submit that: 1) network management activities are permitted by tariff, 2) there is no unjust discrimination, 3) there was no breach of the Company’s notification of network change obligations, 4) the Company is compliant with its privacy obligations and 5) the Company does not control the content or influence the meaning or purpose of communications.

13.Finally, the Company explains why the Commission must avoid unwarranted policy making or regulatory measures which could hurt innovation, impede competition and decrease the efficiency of Canadian telecommunications. The Company will end with a cautionary note to the Commission.

14.The Company further notes that PIAC has submitted an application for costs. The Company notes that the cost claim is premature and cannot be addressed under section 44 of the Telecommunications Rules of Procedure until the record has closed. Further, PIAC has not filed any of the supporting documentation necessary for the determination of a costs claim. Should a cost claim be properly made under the Telecommunications Rules of Procedure at a future date, the Company reserves the right to make submissions with respect to PIAC’s eligibility, the quantum claimed and proper costs respondents.

4.0 A Description of GAS and the Company’s network

15.At the outset, Bell Canada must correct three specific assumptions about its GAS service that emerge from the numerous allegations made by various parties including CAIP: 1) the notion that GAS services are offered over a separate and distinct network than the Company’s retail Internet access service; 2) the notion that a specific amount of bandwidth must be available to GAS customers when tariff rates are paid; and 3) the notion that there are simply no viable alternatives for GAS customers.

4.1 GAS services are not offered over a separate and distinct network

16.The Company’s GAS Tariff9 provides a broadband access service based on asynchronous digital subscriber line (ADSL) technology and enables a service provider to establish a high speed data access path between its end-user’s premises and a Company serving wire centre. GAS uses available bandwidth above the voice-band on the same local loop as the end-user’s Company or CLEC provided residential or business individual line. It includes logical paths to provide network connectivity between GAS Access arrangements and an ADSL Aggregated High Speed Service Provider Interface (AHSSPI) by aggregating traffic associated with each GAS Access served from groups of wire centres to a broadband access server (BAS) and subsequently aggregating such traffic from all Company provided BAS to the ADSL AHSSPI. Essentially, it also includes the backbone to transit traffic from the central office where the end-user is located to the central office with the ISP’s AHSSPI (typically nearest the ISP’s point-of-presence).

17.At least one party10 to this proceeding believed that Sympatico retail Internet and GAS services are offered over separate and distinct networks or at least are isolated from one another. This is not the case. The GAS service offered to wholesale ISPs and the retail Sympatico Internet access service offered by Bell Canada share and have always shared the same access network and therefore will experience the same benefits and the same problems. As discussed in the Interim Answer, wholesale traffic from all of the ISPs who use GAS and Bell Canada’s retail Internet traffic traverse the same central office facilities, the same access network and part of the same backbone network. Figure 1 illustrates how GAS and Bell Sympatico retail traffic transit through the same path/equipment from the DSLAM up to the BAS (and the associated DPI), the point in the network where retail and wholesale traffic is aggregated. The GAS and retail traffic also transits through some of the same backbone network links since the wholesale traffic must be sent to the wholesale equipment/facilities before going to the Internet. It is in fact impossible to distinguish traffic from a retail Sympatico end-user from the traffic of a GAS end-user until the two services reach the BAS.

Figure 1

18.The Company offers another wholesale high speed Internet service to ISPs called High Speed Access (HSA). The Company’s HSA Tariff11 is based on the same ADSL technology as for the GAS service. It enables a service provider to establish a high speed data access path between its end-user’s premises and a Company serving wire centre, and it uses available bandwidth above the voice-band on the same local loop as the end-user’s Company or CLEC provided residential or business individual line. However, the similarities with GAS service end there. The primary distinction is that while HSA traffic transits through the same access network as GAS, HSA accesses are not aggregated via a BAS. The HSA service includes a dedicated Permanent Virtual Circuit (PVC) between the end-user’s premises and the AHSSPI located in the Company’s wire centre. This dedicated channel can be used by the ISP to deliver features that require a permanent IP address, such as monitoring. The HSA traffic does not pass through the DPI devices and is therefore not subject to Bell’s traffic shaping.

No specific amount of bandwidth must be available when tariff rates are paid

19.According to CAIP and Vaxination: i) “The purpose of GAS is to create a pipe or pathway that runs from the premises of each end-user customer through Bell’s central offices (COs) and then on to a physical interface point in Bell’s local network where competitors must interconnect in order to gain access to their customer’s traffic”12, and ii) “A service provider pays a GAS/HSA for each individual ADSL line. The service provider should have full and unqualified access to the bandwidth of each ADSL line the service provider pays for. A service provider also buys sufficient capacity in the “Aggregated High-Speed Service Provider Interface” (AHSSPI) to support the link between itself and the Bell ADSL cloud. Bell is therefore already compensated on the number of customers as well as the total bandwidth generated by those customers.”13

20.Based on these premises, parties throughout their submissions compare the theoretical speed of the GAS service with the reduced speed applicable to a single type of application (i.e., P2P file sharing applications) during peak periods when traffic shaping takes place. These parties then draw the ridiculous conclusion that Bell “cripples the ADSL by approximately 90%”14 or “Since Bell throttles downstream speeds to 30 KB/s (or 240 kbps)15, this represents only 4.8% of the Basic - Residence downstream speed of 5 Mbps referenced in Bell’s GAS tariff.”16

21.The incorrect premises lead to invalid calculations and comparisons, which only serves to confuse the relevant issues. At the outset, because of the nature of DSL services, it is impossible for the Company to promise a specific bandwidth to high-speed service end-users (retail or wholesale GAS). Therefore, these services are marketed with an “up to” speed (such as 5 Mbps in the case of GAS). This is a fact that all ISPs are aware of, including CAIP:

“… this service can be affected by the quality of the copper loop facilities that link each customer premise to a Bell host or remote switch. For example, it may not be possible to achieve GAS speeds of 5 Mbps downstream/800 Kbps upstream in the case of an end-user customer that is served by a copper loop facility that has excessive bridge taps or which is approaching the range of the ADSL equipment located at Bell’s host or remote switches.” 17

22.Furthermore, the traffic shaping that Bell applies is limited to P2P file sharing applications during peak periods, and does not impact the speed for other applications such as web browsing or video streaming such that simultaneous use of other non P2P file sharing applications is not affected. An end-user using multiple applications during peak periods should only see shaping of its speed associated with P2P file sharing applications but not other applications. Bell Canada has performed repeated tests during shaping periods and established that wholesale and retail end-users were able to achieve their full bandwidth tier with non P2P file sharing applications. Therefore, it is inappropriate for CAIP to perpetuate the myth that end-users no longer have access to their full speed or bandwidth.

23.CAIP and other parties make similar arguments regarding the AHSSPI. For instance, CAIP states that “competitors are now paying for AHSSPIs that they either no longer need or which are sized too large because Bell is deliberately reducing the volumes of traffic that it delivers to competitors at the AHSSPI.”18

24.In one of its interrogatory responses19, CAIP describes the factors used by its members to determine the capacity requirements in order to select the appropriate AHSSPI size. These factors include historic usage and growth rate in the customer base. No doubt these factors play an important role for planning purposes, but CAIP itself admits that “the AHSSPI must be sized to peak period utilization by customers, not average utilization”20. It is useful to note that Bell’s 400 Mbps and 1000 Mbps (1 GigE) AHSSPIs are offered at the same monthly rate of $1,850 and its 100 Mbps AHSSPI is offered at $775 per month. Thus, it is generally more economical to subscribe to a 1 GigE AHSSPI than it would be for multiple 100 Mbps AHSSPIs. GAS customers have taken advantage of this situation as evidenced by the fact that many ISPs already subscribe to multiple GigE AHSSPIs. Competitors will likely continue to select an AHSSPI based on the lower tariff rates for GigE links and use the peak utilization as an additional factor in their decision process irrespective of fluctuations in traffic.

25.In any event, contrary to CAIP’s allegations, Bell did not observe any significant drop in peak traffic utilization on the AHSSPIs. For comparison purposes, Figure 2 below illustrates the peak downstream traffic at individual AHSSPIs when Bell compared i) the traffic from 4 March 2008 prior to the implementation of wholesale traffic shaping with ii) the traffic from 3 June 2008 after the implementation of traffic shaping. For example, the analysis shows in the case of one significant GAS customer, #, that the traffic kept growing even after traffic shaping and an additional GigE link had to be added.

26.In addition, CAIP notes21 that its “members typically plan for additional capacity when peak AHSSPI utilisation rates hover between 50 and 60 percent of the total capacity of the AHSSPI”. The table shows that even after shaping the utilization level surpasses the level where additional capacity should be planned for. Therefore, one can see that at this level of utilization, the GAS customer is fully utilizing its AHSSPIs. The utilization levels for each GigE AHSSPI greatly surpass those that CAIP notes are typically used to determine when to add capacity.

Figure 2

27.This GAS customer is definitely not paying for excess capacity.

There are alternatives for GAS customers

28.CAIP brings up a further point of contention when it denies22 that there are viable alternatives to GAS for competitors which, if taken advantage of by independent ISPs, would completely avoid the network management initiative to which GAS is subject. CAIP claims it does not view co-locating in COs and installing their own DSLAMs as a feasible option mainly because of the high utilization of remotes to provide ADSL capable lines in Ontario and Québec. CAIP states that it is its “understanding” that between 40% and 80% of lines are served by remotes and this would prevent them from offering high speed services. The Company provided the number of lines connected to its DSLAMs in confidence to the Commission in an interrogatory response23 in another proceeding and can state that the numbers provided here by CAIP are grossly exaggerated. Notwithstanding this exaggeration, it is incorrect for CAIP to assume that the presence of a remote automatically prevents a competitor from leasing unbundled loops. In locations where Sympatico retail high-speed end-users are served from remote DSLAMs, competitors can still request unbundled loops for DSL and in many cases the Company is able to provision such a loop.24

29.While CAIP agrees that HSA could be a “partial” alternative to GAS for business customers, it denies that third party Internet access (TPIA) service could be an alternative because of service coverage and technical problems. While making this allegation, CAIP merely refers the reader to a previous filing25 where the so-called coverage and technical problems are barely described. The three problems CAIP finds with TPIA are: i) the limited coverage of cable network in business areas, ii) the inability of TPIA service to provide dedicated channels in order to offer static IP addresses, and iii) the fact TPIA is not offered in an unbundled fashion. If CAIP believes there are issues with TPIA, then it should deal with these in a separate application as TPIA is not relevant to the Company’s network management practices. Nevertheless, with respect to the coverage issue, at the end of 2005, cable BDUs passed 97%26 of households and of those passed, 89% had broadband capability. Bell Canada also estimated that in Ontario and Québec, 95% of the SMB customers were in the footprint of a cable operator27. Therefore, coverage is not a serious issue at the very least for residential and SMB markets where most GAS services are prevalent. In terms of dedicated channels in order to provide static IP addresses, this is not a feature provided with GAS service either. Rather, this feature is offered with the HSA service. As mentioned above, CAIP is in agreement that

HSA is an alternative to business end-users. Finally, with respect to unbundling, Bell points out that GAS, even though multiple items are listed in the Company’s GAS tariff, GAS service is also only offered in a bundled fashion.

30.In conclusion, the Company submits that there are a number of viable alternatives to GAS service, and CAIP has not provided credible arguments to the contrary. In fact, due to these alternatives, the Commission has classified GAS service as a non-essential service.28

5.0 Technology description

31.Many misconceptions surrounding the Company’s network management practices seem to result from a misunderstanding of the underlying technology. In order for readers to fully understand the issues, the Company will provide a description of the following: how traffic is routed on the Internet, how devices and applications on the Internet handle congestion, and how P2P works.

How is traffic routed on the Internet

32.Internet applications, such as email and web browsers, are used to access and communicate between parties (servers or other end-users/machines) over the Internet. These Internet applications use well defined protocols to communicate between parties. In order to get data (i.e., content) to another party on the Internet, the application will first format the data and header information (e.g. the recipient’s email address in the case of an email application) will be added. The original data (i.e. content) with its application header will be sent to the network driver which will also encapsulate this data and add yet another header. One of these encapsulation types is known as the Transmission Control Protocol/Internet Protocol suite (TCP/IP). The TCP header (which contains information such as port number and sequence number) is added and finally, the IP header will be added, which includes information such as source and destination IP addresses. This is depicted below in Figure 3.

Figure 3

33.The TCP protocol is designed to ensure delivery of the data unless the communication path between the two end-points has completely failed. When congestion is important, the TCP protocol still attempts to deliver or receive packets. However, after waiting for a certain period of time for information, the applications may time out and stop responding. TCP is also responsible for congestion detection and avoidance and thus will try to use as much bandwidth as the communication path can offer, without dropping packets. The IP protocol defines a set of identifiers or header information, such as IP addresses, that would allow the network to appropriately route the data traffic (packets) to the intended recipients. Network equipment deployed in an ISP network, typically called a router, examines IP protocol header information, primarily the destination IP address, of each data packet and makes the necessary routing decision to forward the data packet towards the receiver. These routers are also designed to efficiently route the traffic, handling failures by re-routing around parts of the network that have experienced communication failures.

34.As indicated in Figure 3, in addition to the TCP and IP protocols, each Internet application supports its own application specific protocols. For example, email applications will conform to specific protocols that control how email should be delivered and received. Web browsers and web servers will conform to specific protocols that describe how web pages will be accessed and how to communicate with the webserver. P2P file sharing protocols (such as BitTorrent) will specify how file transfers should be initiated, carried out and terminated. These application specific protocols are described via a set of protocol headers that are typically transferred at the initial setup phase of the communication between the two end-points. To use the postal analogy, these protocol headers are the digital equivalents to the address label, content declaration (when the application port is properly represented) and postage on a package handled by the postal system.

35.To exchange content, the sender application will typically first exchange a set of “setup” messages with the receiver before the actual data or content will be exchanged. In the case of P2P file sharing, a set of messages is also exchanged before the data or content is shared.

5.2 How do devices and applications on the Internet handle congestion?

36.Essentially, congestion occurs when a link or device along a network path receives more traffic than it can transport to its destination. Internet routers deal with congestion by simply removing or dropping excess packets that cannot be delivered. In addition to lost packets, congestion may lead to latency (packet transmission delivery delays) or jitter (varying delays in packet transmissions). Lost or delayed packets, therefore, are an indication of congestion in a network. Whereas it is easy to understand the importance of minimizing packet loss and latency, jitter is often forgotten or misunderstood. John Bartlett of Netforecast Inc. provides a clear and helpful explanation:

To define jitter we need to back up and understand latency. Latency is the time it takes a packet to traverse the network from source to destination. Latency is half of the ‘ping’ time, or half of the round-trip delay. Jitter is the variation in latency. For example, suppose that the average latency from source to destination is 100 milliseconds (ms). If a specific packet, packet A, traverses the network in exactly 100 ms it arrives just as expected and has a jitter of 0. If packet B traverses the network and is delayed slightly, arriving 130 ms after it left, it is 30 ms later than expected, and has a jitter of 30 ms.

When we send real-time traffic across the network we are trying to encode and then decode a real-time event such as sound or a visual image. The sound and the visual image change constantly, so we have to continually take samples, encode them, send them across the network, decode them and reproduce the sound and stuff on the far end. The receiving end is expecting a continuous stream of data and needs that data to arrive at regular intervals so it can properly recreate the original audio or video. If a packet is late, the time slot in which that data was needed has gone by and the arriving packet is of no use.

Because we know that IP networks are asynchronous and can cause delays in the packets, we implement a jitter buffer on the receiving end. Let’s consider a 40 ms jitter buffer. The jitter buffer predicts the expected time of arrival for each packet, but then delays the playing of those packets by 40 ms. So the real-time event at the far end is being recreated 40 ms later than it could otherwise have been recreated. The value of this is that if a packet arrives less than 40ms late it can be pushed ahead in the jitter buffer so that it is available for its play window even though it arrived late. This is like your colleague telling you the train leaves a half hour earlier than it really does because he knows you often arrive late. The train really leaves a half hour later (the jitter buffer) than the expected schedule.

So a 40 ms jitter buffer will take care of network jitter up to 40ms. If packets are later than that, then again their play window has gone by and they are discarded by the jitter buffer. So why don’t we just make the jitter buffer arbitrarily long to allow for any amount of jitter in the network? Remember that delay. As we delay the recreation of the voice or video image, we reduce the ability for participants to easily interact. When there is a delay on the connection we find ourselves stepping on each other’s speech. This effect is disconcerting. It can make you wonder if the other party is listening, and it can make a back and forth discussion very difficult. So we limit the size of the jitter buffer. This means we need to ensure that the network can keep packets within the jitter specification that the jitter buffer can handle.29

37.Therefore, different web applications have, by necessity, different Quality of Service (QoS) requirements. As mentioned above, latency or jitter can make Voice over Internet Protocol (VoIP) conversations disconcerting and difficult. However, that same latency or jitter will have less effect on file transfers, especially when the transfers are over TCP, since TCP will retransmit lost packets and control latency issues.

38.To illustrate, various web applications or types of traffic can essentially be classified into 3 main categories: Real-Time, Time-Sensitive and Non Time-Sensitive. Real-time applications such as VoIP, live video conferencing and interactive/online gaming must be delivered with minimal latency, jitter or loss for a good user experience. For Time-Sensitive applications, such as HTTP video streaming (e.g., YouTube), streaming audio, web browsing or time-sensitive P2P applications like Joost, minimal packet loss or minimal jitter will not overly impact the end-user experience, however, the end-user is expecting a timely and consistent response. Many applications in this category have some form of buffering; for example, an application may buffer 10 seconds before starting a video, and thus can afford up to 10 seconds of lost data without the end-user noticing. Finally, Non Time-Sensitive applications such as P2P file sharing or large file server downloads are unlikely to be affected by moderate packet loss, jitter or latency since these applications normally have inherent mechanisms that address reorganization of delayed packets and retransmission of lost packets.

39.TCP has congestion control algorithms that make sure packets are sent as fast as possible. All applications that utilize TCP for data transmission incorporate the use of a congestion detection and avoidance algorithm known as TCP Windowing. The goal of TCP is to maximize throughput. Each individual TCP session will try to use as much bandwidth as the communication path can offer without dropping packets. When TCP detects congestion it will scale back the transmission rate. Since each individual TCP session behaves and uses the same scale-back algorithm, each session will get a fair share of the bandwidth. This seems fair for applications that behave the same way on the network. But what happens when applications behave differently, such as P2P?

5.3 What is P2P and how does it work?

40.The term P2P is generally used to describe a communication architecture that has multiple communication nodes, with each node functioning as a “peer” node, in which it can act as both a client and a server. Unlike more traditional client-server architectures, where clients connect directly to a specific server (for example, web browsers are clients that communicate to a specific HTTP server), a network based on P2P architecture is self-scaling; that is, the more nodes that join the network, the more resources there are and resiliency is improved. More and more applications are transitioning away from client-server architectures to that of P2P. For example, some Internet applications are based on a P2P network architecture, such as Skype (a VoIP service) or Joost (an online video streaming service). As each user joins Skype or Joost, they essentially bring along more storage, more bandwidth and more resiliency. As such, by making use of P2P architectures many application service providers can avoid centralized server and network infrastructure investments.30

41.P2P networking does not have the notion of separate client and server nodes (computers), but only equal peer nodes that can make all nodes act simultaneously as both “clients” and “servers”. Nodes can connect to other nodes to receive data and simultaneously serve data to other nodes. When a P2P application attempts to download data, it will create numerous connections to other nodes in order to download pieces of the data from multiple end nodes, reassembling the data upon successful receipt of all the pieces. In addition, once a piece of data is downloaded, P2P applications typically begin to share that piece with other nodes. The emergence of P2P applications on the Internet has created a dramatic change in network traffic behaviour, moving from a predictable interactive “query-response” traffic behaviour to an “always on, multi-query multi-response” traffic behaviour utilizing as much network capacity as possible.

42.One of the principal applications for P2P is file sharing, because each node in the network that wants a file can download and store the content (or pieces of the content) and act as a server for the other nodes in the network. As noted above, P2P applications used for voice and streaming tend to be more real-time than P2P file sharing applications. Users of Skype, for example, engage in real-time voice calls or chatting. In addition, P2P voice communication uses very little bandwidth compared to P2P file sharing applications. P2P file sharing is typically used to download or upload large files (and often many files simultaneously) and tends to be non real-time. End-users typically select files to download, and let these download in the background and often overnight or many days until the file is complete. It is typical to see a large percentage of network bandwidth consumed by file sharing applications. Because P2P file sharing applications are non time-sensitive, these can be “slowed down” during peak hours of Internet use without interrupting the user’s Internet service.

43.In contrast, typical interactive applications, such as web browsing, send and receive a certain amount of data every so often, but most of the time the computer is waiting for user interaction. The unattended P2P file sharing applications, on the other hand, are designed to use as much bandwidth as possible and continuously strive to auto-correct for best performance.

44.There are three ways that P2P file sharing applications use a disproportionate amount of bandwidth compared to other non P2P file sharing applications. First, as stated by Rogers’ Chief Strategist, Mike Lee, P2P is “actually designed to overwhelm other traffic.”31 A P2P application, rather than opening up only one stream or session, will open up 40 to 100 TCP sessions in an effort to transfer data as fast as possible using multiple sources and can therefore grab dozens to 100s times more bandwidth than a traditional single-stream application such as email or Internet banking applications (see Figure 4 below). By initiating more and more P2P applications on powerful computers, the user will continue to expand the number of active streams eventually consuming all available bandwidth. To further compound the bandwidth demand, some users will employ multiple computers on the same Internet connection.

Figure 4

45.Second, once all the available bandwidth is being consumed, TCP, through its windowing technique mentioned above will use a queuing technique for additional requests until more bandwidth becomes available. In addition, P2P applications typically have slots for the total number of downloads. Once the total number of slots is consumed, the P2P application will use a queuing technique for additional file downloads until additional slots become available. The P2P application queuing of multiple requests combined with TCP windowing and the inherent application persistence of P2P enable it to sustain a continuous maximum network traffic load, 24 hours a day, 7 days a week and 365 days a year, as long as there are queued requests.

46.Finally, it is not enough to simply manage the traffic of the network’s users. Some P2P file sharing applications constantly look for the fastest node available, and thus any increase in capacity to one network node will attract increased P2P file sharing upload requests from other P2P file sharing applications resident on other networks.32 As described by Rogers’ Chief Strategist at the latest Telecom Summit, Rogers’ tests have indicated that an increase of capacity at a node could be eaten up by P2P file sharing applications within 24 hours. Indeed, the Company’s own testing shows that in some cases the increase in capacity could be eaten up in as little as 30 minutes.

47.Thus, P2P file sharing applications not only generate considerably more traffic volume in the long run because it runs for a lengthy amount of time, it also uses up the major share of the available bandwidth at any given point in time because it uses so many concurrent TCP sessions and some are designed to constantly look for nodes with the most capacity. Therefore, hundreds of TCP sessions generated by a single user, or users from other networks, will negatively impact the experience of many others. This is compounded by the fact that P2P file sharing users tend to initiate multiple large file downloads and simply leave the application downloading and/or uploading pieces of content. As the content is unaffected by slowdowns, a slowing down of certain of these bandwidth hungry applications (i.e., P2P file sharing applications) has the least impact compared to other types of traffic on the web.

48.There seems to be a belief among certain parties that this movement away from a centralized client-server network architecture infrastructure means that all P2P applications are operated at zero cost. For example, Google states that P2P applications “enable content distributors to massively reduce costs”.33 This belief tends to be coupled with the idea that ISP network management destroys innovation. However, these views are misguided. P2P file sharing does not diminish costs, rather it offloads the hosting and server costs from a content provider to the ISP’s network. In other words, P2P file sharing creates a negative externality for facilities based ISPs. One could argue that content providers need to be motivated to develop a more efficient method of distribution that takes into account network operator concerns. Network management does not harm innovation; to the contrary, it is a market stabilizer that encourages innovation of new, more efficient applications.

49.Indeed, additional capacity cannot, on its own, resolve this issue. As stated by Mike Lee, “file sharing programs are designed to constantly look for more optimal sources for their shared files, sending out constant requests to find a better connection or more sources for files. They will also keep downloading bits of files from wherever they can, all the while accommodating requests from other file sharing users. If the P2P software senses there is more space on a network, it will flood that space with these requests.”34 The need for bandwidth management is therefore clear. As mentioned by Cisco in its 12 June 2008 comments: “Even if more bandwidth could be added overnight, peer-to-peer (”P2P”) applications are currently designed to use all of the bandwidth that is available. Thus, network management is the only realistic response to the current challenge presented by P2P and other bandwidth-intensive applications.”

6.0 Growth in internet demand

50.Internet demand has grown exponentially both globally and for Bell Canada. The Company will describe the global growth of Internet demand and network congestion. The Company will also provide a description of its Internet demand growth and experienced network congestion.

Global Internet Demand

51.Cisco predicts that total global IP traffic will increase by a factor of six from 2007-2012.35 Figure 5 shows that globally between 2007 and 2012: (i) web, email, file transfer traffic will increase 334%; (ii) P2P traffic will increase 285%; and (iii) Internet video to PC traffic will increase 680%.

Figure 5

52.The Minnesota Internet Traffic Studies (MINTS) also examines Internet growth trends and finds Internet traffic volumes consistent with the Cisco white papers.36 MINTS finds that there was a regular doubling of Internet traffic in the early 1990s, but with a period of ten times growth in 1995-1996 due to an increase in graphics-rich web traffic replacing text-oriented material.37 Around 1997, while growth rates declined, they were still doubling every year and may now be at the point where Internet traffic is growing at fifty percent a year.38 However, MINTS states that four to six times annual growth is still possible since there are potential sources that can drive this growth such as data transfer and video transmission.

53.A recent report by Bret Swanson and George Gilder (Swanson and Gilder) for the Discovery Institute titled Estimating the Exaflood: The Impact of Video and Rich Media on the Internet – A “zettabyte” by 2015?39, discusses a number of sources that will drive significant increases in demand. One of the key drivers identified by Swanson and Gilder is video. For example, by mid-2007 YouTube was streaming 600 petabytes (PB)40 per year, but with the introduction of High-Definition (Hi-Def) videos, a Hi-Def YouTube would stream 12 exabytes (EB) per year or close to the total traffic on the U.S. Internet in 2007. Similarly there is amateur video. Currently, amateurs account for 10 EB of video a year, which can be shared between family and friends, but as more content becomes HD, this would grow to 100 EB per year or 10 times current U.S. Internet traffic.

54.On a more professional basis there is motion picture distribution and Internet TV. Companies such as Amazon, Netflix, Blockbuster and Apple all provide movie downloads, and again with HD, Netflix alone would currently ship 5.8 EB of DVDs each year. For Internet TV, NBC streamed 50 million of its shows on-line in October 2007, and in 2007 Hulu was founded by NBC Universal and News Corp., which now consists of a large selection of videos from more than 50 content providers, including FOX, NBC Universal, MGM, Sony Pictures Television, Warner Bros.41

55.However, video is not strictly confined to entertainment. There also exists video conferencing. By mid-2007, MSN Video Messenger was generating 4 PB per month, but with Cisco’s new HD Telepresence system, just 75 one-hour conferences would generate as much traffic as the entire internet in 1990.

56.In March 2008, comScore found that the number of online videos viewed in the U.S. increased 13 percent to 11.5 billion.42 There were also 138.5 million unique viewers with each viewer watching an average of 82.8 videos.43 comScore also found that: (i) 73.7 percent of the total U.S. Internet audience viewed online video; (ii) the average online video duration was 2.8 minutes; and (iii) the average online video viewer watched 235 minutes of video.44

57.As shown by Cisco in Figure 6 below, YouTube traffic alone generated more traffic in 2008 than the entire U.S. backbone in 2000.45

Figure 6

58.Another key driver identified by Swanson and Gilder is imaging and data gathering transmitters. For example with imaging, in 2006 there were 400 million camera phones sold, and in 2009 it will increase to 700 million with each photo generating files of around 3 MB, and similarly with the introduction of Apple’s iPhone, video rich content can be easily obtained for mobile devices. From a business perspective there is Radio Frequency Identification. While the total impact is not known, in 2000 there were about 100 million devices connected to the Internet, but by 2010 that number is expected to be 14 billion. This exponential increase will be facilitated by Internet Protocol version 6 (”IPv6″) which will allow for 3.4 x 1038th unique addresses, allowing for even more devices to be connected to the Internet.46

59.The final key driver identified by Swanson and Gilder is the move away from Local Area Networks to “cloud” computing. Network computing and web services or “cloud” computing presents a true virtual computing environment, allowing use of web service interfaces to requisition machines for use, load them with custom applications, manage networks’ access permissions, and run as many or few systems as required. Cloud computing is currently being used. Since cloud computing delivers software and services over the Web and through a browser, it is no longer necessary to install servers and client software.47 For example, Amazon Web Services offers a web-scale computing platform which allows developers and businesses to scale computing infrastructure up and down depending on immediate requirements.48 Currently, over 330,000 developers, start-ups and established companies are using Amazon Web Services. Other cloud computing providers include IBM Blue Cloud49, GoGrid,50 3tera,51 and Microsoft Live Mesh.52

60.There is also the storage and transmission of digital information. IDC estimates that the total digital information created, captured, or replicated in 2006 was 161 EB, but that by 2010 it could be 988 EB, and if Internet content companies continue to offer remote or online back-up services, then by 2015 there could be a total of 50 EB of remote back-up in the U.S. alone.53 For example, the New York Times uses Amazon Web Services to digitally archive issues of the New York Times starting with Volume 1, Number 1 September 18, 1851 and currently ending at December 30, 1922.54 Thus, subscribers now have access to these archived issues of the New York Times which are stored and accessed over the Internet.

61.Similar to other ISPs, Bell Canada is experiencing growth in utilization of the network by existing customers. In 2002, the average Bell Canada retail Internet customer used # Gigabytes (GB) of bandwidth per month. By 2008 this number has grown tenfold to # GB per month. As Figure 7 below demonstrates (in Gigabits per second), the overall impact on the network has been a huge growth in demand.55 Although some of this demand is partly a result of subscriber growth, the number of retail Internet subscribers for Bell Canada over the last year has largely remained flat and thus these growth figures can only be partially attributed to subscriber growth.

Figure 7

7.0 Increased network congestion

7.1 Congestion: A Global Issue

62.As noted above, most applications that run over the Internet use protocols with inherent congestion detection and control mechanisms that are able to automatically adapt the delivery speed of packets when congestion is detected. However, these mechanisms are effective only to the extent that traffic sent across the network does not exceed the capacity that each of the network elements are capable of processing at any given moment. When capacity is exceeded, packets are delayed before being delivered to their destination (also called latency) or dropped, in which case the user or application may decide to retransmit the packets. Therefore, latency and dropped packets are considered symptoms of congestion in a network. In the broadest of terms, network congestion simply implies that user demand is greater than network capacity. The impact of congestion on Internet end-users will vary from slower traffic to not being able to receive content from the Internet at all.

63.The nature of the growth of Internet traffic is that as network capacity expands, new user applications invariably also grow to utilize that capacity. As a result, Internet service providers have begun to manage network data flow through the use of various network management tools in order to mitigate network congestion due to increased traffic by consumers, particularly during peak periods. The issue of increasing network congestion is not one that is faced solely by Bell Canada. It is an issue that is faced by network providers around the world: as a Comcast executive was recently quoted as saying, “You can never build your way out of this problem.” 56

64.The reason behind network congestion is the dramatic increase in demand relative to capacity. Nemertes Research (Nemertes) conducted an in-depth analysis of Internet and IP infrastructure (capacity) as well as current and projected traffic (demand) in order to determine if there will ever be a point where demand exceeds capacity.57 The Nemertes Study examines demand independent of capacity in order to control for the situation where capacity may be limiting demand.58 The Nemertes approach was to focus on the absolute busy hour characteristics of a user group rather than trying to determine a comprehensive user demand profile over time.59 Figure 8 below shows the results of the Nemertes model of user demand for the U.S. By 2012 demand would reach 200,000 PB which is the same as an individual user consuming or generating 26 GB per day.60 While the results of the Nemertes model yield geometric growth in demand, it should be noted that the model also states that in 2006, an individual user consumed or generated about 350 MB per day which is equivalent to downloading an hour of Internet video or multiple hours of working, emailing, uploading, downloading and watching video.61 It is noteworthy that 350 MB per day translates into approximately 10.5 GB per month.

Figure 8

65.Swanson and Gilder agree with the approach taken by Nemertes:62

Any long-range forecast in the dynamic world of technology exposes itself to charges of “speculation.” Yet Nemertes’ approach is conceptually appealing for the very reason that most observers, unable to grasp the power of exponential change, have consistently underestimated the mid- to long-range pace of advance in the world of computation and communication. Professor Odlyzko believes Nemertes’ projections of traffic (or what it calls “demand”) are very aggressive but not beyond plausibility. We agree.

66.On the supply side, from 2007 forward, Nemertes projects that connectivity capacity will increase at a faster rate than core capacity as network providers extend higher speed interfaces toward the edge of the network, and that global broadband access capacity will grow at a linear rate over time.63 This result assumes that global fibre to the home will be increasing over time, and that wireless devices will increasingly replace fixed access technologies.64 However, as technologies in wireless devices improve and data transmission rates to basic devices increases, it is possible that global access capacity could increase more than projected.65

67.Then by comparing demand against supply, Nemertes finds that if innovation continues to drive demand, that absolute demand will approach the limits of capacity by 2012.66 In practice, this means that the rate of retransmission goes up, and net throughput goes down.67 As a result, in order to avoid this problem and close the gap between desired capacity and available capacity, by 2010 the additional investment in just North America would amount to $43 billion. This additional amount is equivalent to nearly 60% of the existing projected network investment of $72 billion by 2010.68

68.The relationship between supply and demand is also noted by Jason Kowal in a report titled The Never-Ending Rush Hour: Internet Traffic Growth Requires Continual Investment in Capacity and Innovation in Network Management:69

Although Internet capacity is upgraded to meet new traffic requirements, the increase in capacity and capabilities also stimulates new traffic growth as well. As described above, aggregate increases in available network capacity supply can also create demand in the network by making new types of content delivery possible. This occurs because there is less contention for resources on end-to-end connection. As a result, Internet traffic growth typically moves in relation to the growth of capacity. However the relationship does typically move in lock-step: in some years traffic grows faster than new capacity is deployed. According to TeleGeography, in 2006 the growth in average traffic level (75%) outpaced the growth of capacity (47%) on the world’s international Internet backbones for the third consecutive year. Significantly, this data only captures the early stages of surging video traffic, which is putting new strains on conventional capacity upgrade processes. [footnote omitted]

69.As Nemertes notes, the implication of this capacity crunch is not that the Internet will stop working, but rather the degradation of Internet performance will constrain users’ ability to consume more.70 If network providers do not take steps to mitigate the congestion issue, users will experience erratic performance (i.e., an increase in latency and dropped packets). Thus, Nemertes believes that congested infrastructure will slow down the pace of both technical and business innovation such that the next Google, YouTube, or Amazon might not arise due to the inability to fulfill that demand.71

7.2 Congestion in Bell Canada’s Network

70.It is well accepted that the Internet consists of multiple interconnected networks and that applications that run over the Internet can experience congestion in locations that may be external to Bell Canada’s network. This submission, therefore, only addresses congestion in Bell Canada’s network as it pertains to the delivery of high-speed Internet services (retail and wholesale GAS). Bell Canada measures latency and dropped packets (referred to as “cells” in the case of an ATM network) as well as the level of utilization of links in the network. Although both measurements are important, it is the level of utilization of links that Bell Canada uses as the primary criteria to determine where the congestion is occurring. Such an approach is more efficient since utilization level measurements are more consistently and readily available than latency measurements. The approach is also reliable given the observed close relationship between the measured utilization levels and the latency (the higher the utilization in a link, the higher the latency).

71.A common industry practice in network management is to develop thresholds at which the utilization level in a link has a very high probability of producing negative impacts on end-users. Therefore, it is not feasible to ever utilize a link continuously to its full capacity. The actual thresholds used by Bell Canada were developed by analyzing the plotted latency data for each type of link in its network. Using this method, Bell Canada was able to determine the specific level of utilization of a link at which the number of congestion events (latency and dropped packets) increases dramatically and the events become longer in duration. Link utilization measurements are taken every 15 minutes, however these figures hide the instantaneous traffic peaks that can be higher than actual measured utilization. This is highlighted by the fact that latency and dropped packets occur well below 100% link utilization. Bell Canada has developed thresholds for each type of link using the same methodology:

DS-3: 61%;

OC-3: 84%; and

OC-12 and OC-48: 90%.

72.The fact the thresholds for DS-3 links are lower than they are for OC-48 links is explained in part due to the lower processing capacity of the DS-3 equipment and the longer periods of time traffic is able to remain in queues waiting to be processed, increasing the potential for higher latency.72

73.The Company is able to actively monitor congested links in order to take appropriate action to resolve the problem. For a link to be considered congested today, the threshold must have been exceeded at least once on 5 or more different days of a 14 consecutive day period. Bell Canada believes that this approach to measuring and identifying congestion in its network is consistent with the methodology used by many other service providers throughout North America and the UK.

74.The Company has broken down its network into four categories, namely DSLAM, Aggregation Network, BAS and Backbone Network. The location of those parts of the network, and the links associated with those parts, can be seen in Figure 9.

Figure 9

75.Figure 10 below illustrates the number of congested links per month for the period of March 2007 to May 2008 for each location in the network (i.e., central office DSLAM, Aggregation network, BAS and Backbone network). Figure 11 expresses this same information as a numerical percentage rather than absolute numbers.

Figure 10 (missing)\<