Bell filed its response [.zip] to the Canadian Association Of Internet Providers (CAIP) submission to the Canadian Radio-television and Telecommunications Commission (CRTC) on its throttling practices yesterday, unsurprisingly arguing that its actions are justified and that there is no need to deal with the issue on an emergency basis. Several points stand out from the submission including its non-response to the privacy concerns with deep-packet inspection (it merely says that it does not retain or use the data, but does not deny collecting what could easily be interpreted as personally identifiable information) and its inference that P2P usage could be deemed using a connection as a “server” and therefore outside the boundaries of “fair and proportionate use” under typical ISP terms of use.

Most importantly, however, Bell provides data on its network usage that significantly undermines its claim that P2P usage is causing such havoc with its network that throttling measures that impact 100 percent of its (and some of its competitions’) users are needed. Bell again reiterates that the “problem” lies with 5 percent of its users that are heavy P2P users. Yet that 5 percent apparently uses 33 percent of available bandwidth during peak periods. That is a disproportionate use to be sure, yet it struck me as far lower than might have been expected.

Data from other countries suggest that other countries have a far more disproportionate bandwidth breakdown. For example, Cisco has claimed that 4 percent of Japanese users account for 75 percent of inbound traffic volumes. Another study [PDF] from Korea found that 5 percent of users accounted for about 50 percent of total bandwidth use. Further, a study last fall by Ellacoya Networks, which sells shaping tools, found that P2P is not the dominant use of broadband networks in North America—web traffic is.

This becomes relevant for at least two reasons. First, Bell has been fighting its public relations battle on the premise that P2P is a problem that necessitates throttling practices that inconvenience 100 percent of its users, yet its own data suggests that the problem may not be as severe as first thought. Second, while the Telecom Policy Review Panel recommended a net neutrality provision that accounts for “reasonable technical constraints,” Bell’s current traffic experience may make it more difficult to argue that constraints that affect 100 percent of users are reasonable in light of better alternatives (Bell acknowledges in its submission that Comcast is moving away from this approach) and network usage patterns that do not seem particularly out-of-sorts.