UBB supporters have long pointed to concerns about network congestion as a key rationale for imposing usage based billing measures on their subscribers. The argument is easily understood â€“ network capacity is limited and congestion on the network negatively affects all subscribers. By imposing measures designed to limit network congestion, the small number of â€œheavyâ€ users who contribute most to the network congestion do not adversely affect the large number of light or average users, whose Internet use has a more limited impact on network congestion.
The link between UBB and network congestion dates back over ten years, when the CRTC first approved volume usage rate restrictions for the wholesale service offered by cable Internet providers known as Third Party Internet Access (TPIA). Shaw and Videotron sought approval for volume usage rate restrictions and the Commission approved the request based on the need to â€œensure fair and proportionate use of the service by all end-users.â€
At the time, cable Internet was a shared service and the potential for one end user to affect the service of a neighbouring user was widely recognized. Accordingly, the CRTC established the following policy, conditioned on the cable providers establishing the same volume usage rate restrictions on their own residential customers:
Shaw and VidÃ©otron propose to apply monthly volume usage rate restrictions to end-users. For example, Shaw proposed an $0.80 charge per 100 megabytes per end-user whose monthly downstream traffic exceeds five Gb and a $2 per 100 megabytes charge for end-users with monthly upstream traffic exceeding two Gb. The carriers generally submitted that their cable network is a shared network where the quality of service available to one end-user is a function of other end-users’ usage. Accordingly, they propose traffic thresholds to ensure fair and proportionate use of the service by all end-users. Shaw and VidÃ©otron indicated they would apply the same usage restrictions and charges to their Internet service end-users as they propose to apply in respect of ISPs’ end-users.
The Commission approves on a final basis the volume usage rate restrictions proposed by Shaw and VidÃ©otron. To the extent that these carriers do not apply these volume usage rate restrictions and associated volume thresholds to end-users of their own Internet services, the Commission is of the preliminary view that such action would be contrary to s. 27(2) of the Act.
Rogers and Cogeco have not proposed volume usage rate restrictions. However, the Commission considers it would be appropriate for each of these carriers to adopt, at its option, volume usage rate restrictions and associate volume usage thresholds similar to those proposed by Shaw and VidÃ©otron. In this event, to the extent Rogers or Cogeco do not apply the same restrictions to end-users of its carrier’s own Internet service, the Commission is also of the preliminary view that such action would be contrary to s. 27(2) of the Act.
While cable providers received regulatory support for UBB, there has since been very limited uptake of TPIA, with some cable providers rendering it difficult for independent ISPs to adopt the service.
The path to UBB for telecom providers such as Bell, which offer Digital Subscriber Line (DSL) services, was somewhat different. With retail Internet services forborne in 1999, Bell was free to implement UBB for its retail customers, which it did in 2006.
Shifting UBB to its wholesale customers was more challenging, however, since independent ISPs purchase a regulated service known as Gateway Access Service. The regulated GAS, much like the TPIA service for cable, is not an Internet service but rather a connection between end users and the independent ISP. It is important to note that the actual provision of Internet services comes from the independent ISP, not from carriers such as Bell. Independent ISPs need the GAS in order to reach the end users themselves, since only telecommunications and cable companies have the “last mile” connection to the customer (i.e. the copper telephone line or cable connection to the home). Many countries require some form of open access to this last mile (particularly for telecom) in order to enhance competition among Internet providers.
The Commission’s UBB support as a mechanism to address network congestion was reinforced in the 2009 Internet traffic management practice guidelines as it emphasized its support for economic measures as preferable to â€œtechnical measuresâ€ that restrict users access through technical means such as â€œthrottling.â€ The Commission started from the position that â€œparties generally acknowledged that some traffic management is required to address congestion in order to ensure that all end-users receive acceptable Internet service.â€ Given the need to address network congestion, the Commission noted that:
economic ITMPs would generally not be considered unjustly discriminatory, as they link rates for Internet service to end-user consumption. Economic ITMPs also provide greater transparency to users than technical ITMPs, as they are reflected in monthly bills. Furthermore, these practices match consumer usage with willingness to pay, thus putting users in control and allowing market forces to work.
Indeed, the support for economic traffic management measures was firmly ensconced in the test the Commission established for reasonable traffic management practices. The test includes a requirement on ISPs to â€œexplain why, in the case of a technical ITMP that results in any degree of discrimination or preference, network investment or economic approaches alone would not reasonably address the need and effectively achieve the same purpose as the ITMP.â€
The Carriers on Congestion
Although the CRTC has proceeded on the assumption that UBB is a response to network congestion concerns, there is considerable evidence to the contrary. Indeed, both cable and telecom operators have acknowledged â€“ in both statements and practices â€“ that UBB is at best only loosely related to network congestion. For example, in November 2010, Rogers, Shaw, Videotron, and Cogeco wrote to the CRTC in support of UBB and acknowledged that it could not set pricing on the basis of costs:
UBB charges are first and foremost a mechanism for managing Internet traffic, that is to say an economic ITMP. UBB rates and caps are set at levels where they are expected to have an influence on the traffic patterns of end-users, particularly high consumption end users, otherwise they do not achieve their purpose.
Any attempt to establish rates for wholesale UBB charges based on a narrow assessment of incremental costs is destined to failure, as it would not take into account the broader behavioral impacts that permit the attainment of the underlying carrier’s network management objectives.
Bell has been even more transparent about the profitability of UBB. In a February 2010 conference call with company analysts, CEO George Cope noted:
And, of course, we’re seeing the everâ€”the growthâ€”the usage of the Internet with the Internet sticks. And I think the really, really important thing for investors, which we are obviously following and executing on, is making sure that it’s a usage-driven modelâ€”and that’s the model that we’re pursuing, and at this point, the model that the Canadian market is pursuing. And that’s important to make sure we monetize this significant opportunity for our investors, and at the same time, for our customers from a great service perspective.
In an August 2010 quarterly call, Cope stated:
our data revenue growth was 3.8% for our Residential Services business, particularly driven through an increase in Internet ARPU of 3.3%. And interesting, almost all that increase now coming from usage based billing as the demand for Internet use explodes through the use of video services, and we’re continuing to see an increase in the revenue per customer.
Three months later in November 2010, Cope noted:
our residential services had an excellent revenue quarter from a data perspective, as well, with data revenue growth of 5%, driven principally by the bandwidth usage revenue being up 83% year-over-year.
Later in the same call, Cope linked increased Internet video demands with growing revenues â€“ â€œas we see a growth in video usage on the internet, making sure we’re monetizing that for our shareholders through the bandwidth usage charges.â€
In other words, while the company has claimed that UBB only affects a small percentage of subscribers, it is the primary driver of increased revenues from its Internet access services division.
Shaw has also told analysts that UBB represents an important monetization opportunity closely linked to customer video usage. In a November 2010 presentation, executives stated:
In the future, we believe our usage based billing plan will enable the further monetization of our Internet business as data usage becomes more prevalent and common amongst our customer base (ie. streaming of video)
Competitors have made much the same point. For example, in reviewing Bell’s pricing, Primus stated “it’s an economic disincentive for internet use. It’s not meant to recover costs. In fact these charges that Bell has levied are many, many, many times what it costs to actually deliver it.”
In fact, Bell’s earlier proposal for UBB pricing for the GAS service (now replaced by AVP) similarly demonstrates that it bore little relation to actual costs or network congestion. First, its plan was different in Quebec, where there is a 60 GB cap, and Ontario, where the cap is set at 25 GB. The difference was plainly a function of the competitive environment, where Videotron’s 60 GB cap forced Bell to offer a similar cap in order to remain competitive. Moreover, Bell’s plan featured a 60 GB cap with an overage charge for the next 20 GB. After 80 GB, there was no further cap until the user hits 300 GB. In other words, using 80 GB and 300 GB would have cost the same thing. This suggests that the plan had nothing to do with pay-what-you-use but was rather designed to compete with similar cable ISP data caps.
A Congested Network?
Network congestion is often treated as a catch-all for a network where demand exceeds capacity and therefore cannot provide all users with maximum speeds. The CRTC has defined it as â€œa situation whereby the amount of traffic transiting the network may lead to a deterioration in service for some end-users.â€ A closer examination of incumbent networks reveals that Internet traffic traverses at least three stages, two of which typically do not raise significant congestion concerns.
The first stage is the so-called â€œlast mileâ€, the link between the end-user and the â€œCentral Officeâ€ (CO), where ISPs begin to route user traffic to its intended destination. The last mile is the most coveted part of the network since the ability to serve an end user with Internet services depends on last mile access. While the last mile is a critical component of the network, congestion is limited, particularly for DSL services provided by telecom companies such as Bell. This should come as no surprise since the ISP is able to limit the amount of data carried on the last mile by establishing a bandwidth speed consistent with its network capabilities. Moreover, since DSL services involve a direct connection between the end-user and the CO, there are no other users to congest the network.
The incumbent ISPs have acknowledged the absence of network congestion in the last mile stage. At a recent Standing Committee on Industry, Science and Technology hearing on UBB, Bell’s Chief Regulatory Officer Mirko Bibic acknowledged:
There is a copper loop that goes from our central office to the home and all data travels on that pipe: Internet traffic, television traffic, voice traffic, long-distance traffic. But there are no congestion issues there. The real issue is when you get to the central office and go behind it into the Internet. Fibe TV is completely different.
The second stage runs from the CO until the ISP hands off the user to another provider. This may occur at several different points. In some instances, an independent ISP may co-locate with the incumbent ISP by installing their own equipment at the CO, so that the user only uses the incumbent’s last mile, not their internal network.
In other instances, the incumbent ISP may aggregate the traffic of many users together, delivering the collective traffic to the independent ISP (if the users are customers of an independent ISP) at a later point in the network. Since the ability to deliver faster speeds to users depends in part on shortening the distance between the user and the CO, the investment in the network (which in many urban areas has been built for decades) is focused on a closer connection to the end user. With a close connection to the end user, the traffic is routed along the ISP’s internal network until it reaches the third stage â€“ the Internet.
Once the user traffic reaches the Internet stage, it moves between other network providers to its end destination (e.g. a website or an email in-box). For traffic carried by independent ISPs, this traffic transits separately from Canadian ISPs such as Bell since independent ISPs acquire sufficient connectivity to handle their customers’ traffic demands. For the incumbent ISP traffic that originates from its own residential customers, it too is managed through peering and transit arrangements. The Internet stage naturally involves a global network and there are no significant congestion concerns at this stage either (or at least no congestion concerns specific to Canadian ISPs).
The issue of network congestion in Canada is therefore largely limited to the second stage when the user traffic enters the incumbent ISP’s own internal network (beyond the last mile) until it is handed off to another provider such as an independent ISP, a peering provider, a Content Distribution Network (CDN) provider, or a provider by way of a transit arrangement for Internet services. While congestion can arise within this internal network if the simultaneous aggregated traffic demands exceeds network capacity, there are no additional costs to the ISP â€“ the congested traffic all runs within its own network.
Moreover, there are many ways to address this congestion. In the case of independent ISPs, the most obvious method is to hand off their user traffic closer to the CO so that it does not contribute to the congestion.
For the larger incumbent ISPs, â€œtechnical measuresâ€ â€“ also known as traffic shaping â€“ have become a common practice for addressing congestion issues. Virtually all of the larger Canadian incumbent ISPs use some form of traffic shaping to limit the bandwidth allocated toward high-bandwidth applications during peak periods. The incumbent ISPs apply these measures to all traffic regardless of whether it is their own retail traffic or traffic that originates from independent ISP subscribers.
There is also reason to believe that incumbent ISP networks are large enough to handle the Internet traffic without concern for congestion since the same networks simultaneously carry other high bandwidth traffic such as IPTv. The issue may well be one of bandwidth allocation. If Internet traffic demands continue to grow faster than the other bandwidth demands running on the same connection, carriers could shift some of the â€œspaceâ€ reserved for services such as IPTv (or the chicken roasting channel in the case of cable) to the Internet and thereby relieve some of the congestion pressures.
None of this suggests that consumer broadband demand is not growing rapidly. Driven by increasing use of the network for streaming and downloading video as well as data intensive games, it clearly is. However, the demands on the network are not outside historical norms nor do they necessarily mean that the network is now â€œcongested.â€ Rather, it suggests that certain parts of the network may face greater congestion strain during certain periods in the day, which can be addressed through several mechanisms, including increased investment, technical measures such as traffic shaping, and a re-examination of bandwidth allocation on ISP networks.