Rogers throttling may breach net neutrality rules

Rogers throttles file-sharing traffic from BitTorrent more than any other internet provider in North America and may be violating Canadian net neutrality guidelines, a U.S. researcher says.

Data from connection speed tests run by users show that between the second half of 2008 and the first half of 2010, Rogers slowed down or throttled BitTorrent peer-to-peer traffic 78 to 91 per cent of the time, reports Syracuse University information studies researcher Milton Mueller in an interactive table posted on his website.

"Under the regulations that the CRTC promulgated for reasonable internet traffic management practices, I think 100 per cent, 24/7 throttling is not conformant," Mueller said in an interview Tuesday.

"So I think consumers would have a basis to complain and the CRTC would have a basis to act."

In a statement emailed to CBC News, Rogers said its network management "is in full compliance with CRTC regulations."

According to the company's network management policy, high-speed internet and portable internet customers face a maximum upload speed of 80 kilobits per second at all times for peer-to-peer file sharing. Rogers's advertised upload speed for its fastest internet service is up to two megabits per second or 26 times faster.

Rogers says it limits uploads of high-volume traffic that is not time sensitive, such peer-to-peer file sharing, to ensure a high level of service for time-sensitive tasks such as voice and video applications. It added that it does not limit download speeds for any application or protocol.

Rogers and other ISPs distinguish peer-to-peer traffic from other types of traffic using a technique called deep packet inspection.

Mueller's data was highlighted in a posting by Toronto-based technology blogger Pete Nowak on Oct. 24, the second anniversary of the CRTC's internet traffic management or "net neutrality" rules.

Those rules state that technology to manage internet traffic:

• Must be designed to address "a defined need and nothing more."
• Should be neither "unjustly discriminatory nor unduly preferential."

"The whole rationale for the throttling is you're conserving bandwidth for the people," Mueller said. "And the fact of the matter is, unless your bandwidth is congested 100 per cent of the time – in which case you ought to be maybe expanding it —  there's no reason to be throttling 100 per cent of the time."

Affects 'viability' of some services

He added that throttling targeted at particular applications will affect internet users' choices.

"It's definitely something that's going to affect the viability of different kinds of services."

Data on throttling by ISPs around the world was crowdsourced through a test called Glasnost developed by the Max Planck Institute for Software Systems in Germany. Users visit a website where the test is hosted. It measures and compares traffic speeds for different applications to detect throttling, including whether it is applied to uploads or downloads, certain types of traffic, or traffic linked to certain port numbers.

The results, which include the IP address of the ISP involved, are stored in a database managed by Google M-Labs and organized by date.

The data show that the only other Canadian ISP that throttled BitTorrent connections more than 31 per cent of the time was Cogeco, which reduced its level of throttling from 82 per cent in the first half of 2009 to 44 per cent in the first half of 2010 following consumer complaints. The U.S. showed even lower levels of throttling and a sharper decline, with only one ISP conclusively registering any BitTorrent throttling at all in the first half of 2010 – Clearwire, with just 17 per cent. The data show a "background" of five per cent even if no throttling is taking place, so that only results above that threshold are conclusive.

Mueller is in the process of compiling the data for late 2010 and the first half of 2011.

In a recent study presented at the Telecommunications Policy Research Conference in Arlington, Va., in September, Mueller compared the public and regulatory reaction to the practice of BitTorrent throttling in Canada and the U.S.

U.S.-Canada paradox

"We got this very paradoxical result in which Canada has unambiguous, clear authority to impose net neutrality but they really did a lot less about BitTorrent throttling," he said, noting that the consumer outcry over the throttling practices was notable in both countries.

He suggested that perhaps the Canadian regulator is "a bit too cozy" with large Canadian telecommunications companies.

"I think they do give too much weight to Bell Canada and some of these heavyweight operators," he said. "I think that's starting to change now that they're starting to look more carefully at the use of this throttling."

This past September, the CRTC issued guidelines for resolving consumer complaints about throttling, with timelines for each step. It announced that a summary of complaints would be published four times a year and violators could face a third-party audit or even a public hearing. It also  published a document explaining what ISPs are allowed to do or not allowed to do to manage their traffic and how consumers can make a complaint.

The announcement of the new guidelines came after months of complaints from gamers that Rogers throttles some non-file sharing traffic such as online games. Rogers said it had resolved the problem for World of Warcraft. It has acknowledged that other games and applications may be unintentionally throttled under certain circumstances. A complaint from gamers concerning other games remains before the CRTC.

Mueller said that type of complaint isn't surprising — the technology used to identify peer-to-peer traffic might get confused sometimes. That's one of the reasons why he thinks consumers should choose internet providers who use internet traffic management technology that targets points of congestion rather than singling out particular applications.

Internet providers that use application-specific throttling argue that it's necessary to improve users' experience of online activities such as streaming video that are more noticeably degraded when the network is congested.

"That's in some ways a justifiable argument," Mueller said. "But also, once you start picking out different applications, you're getting into some very difficult, technical discriminations that might have unintended consequences."

Mueller said his future research will examine whether other protocols besides BitTorrent are being throttled by ISPs and will look into the use of deep packet inspection for other purposes such as copyright policing, placing ads, government surveillance and censorship.