Dan Rayburn | Wednesday November 5, 2014
Last week when M-Lab published the results of their study on traffic congestion on the Internet, many used it as another example of how the ISPs were the one’s responsible for slowing down or blocking Netflix traffic coming into the ISPs network. But something odd stood out about some of the data M-Lab’s collected, which policy makers and lawyers missed. The M-Lab’s data documented the introduction of a higher level of prioritization somewhere on the network. Prioritization rules that would impact how content is delivered to the ISPs, and would greatly impact the quality of video and other content consumers would receive.
This morning, Cogent admitted that in February and March of this year the company put in place a procedure that favored traffic on their network, putting a QoS structure in place, based on the type of content being delivered. Without telling anyone, Cogent created at least two priority levels (a ‘fast lane’ and ‘slow lane’), and possibly more, and implemented them at scale in February of this year. What Cogent did is considered a form of network management and was done without them disclosing it, even though it was the direct cause of many of the earlier published congestion charts and all the current debates.
Cogent said they prioritized data based on user type “putting its retail customers in one group and wholesale in another.” Cogent said “retail customers were favored because they tend to use applications, such as VoIP, that are most sensitive to congestion” and that they “implemented a QoS structure that impacts interconnections during the time they are congested.” Cogent classified M-Lab performance tests into the highest priority class. As they did so, this change instantly and dramatically improved the M-Lab test results. [Described by Susan Crawford as "the Cliff and the Slope" – it appears the cliff was Cogent taking on Netflix and the slope was the introduction of a fast lane on the Cogent network] The fact that the high priority traffic class so quickly improved M-Lab’s test performance demonstrates conclusively that there must have been significant congestion present in Cogent’s network. Cogent now admits to impacting third-party content and says they did it as a “last resort effort to help manage the congestion and its impact to our customers”.
It seems extremely unusual that Cogent implemented this traffic change during the very same week that Comcast announced the Netflix deal. The Comcast and Netflix interconnection deal was announced On February 22 and most of the Cogent changes, according to the M-Lab blog, appear to have occurred within five days starting on February 20. Was this done to influence public debate about the cause & effect of the deal? Remember, Cogent started making critical comments in the press once they started making their prioritization changes, while not disclosing what they had done. And to make this even more interesting, Cogent makes it sound like they only did this in February and March of this year, but Cogent’s priority classes appear to continue to this day, which can be proven by looking at how the packets are being prioritized. If Netflix was aware of their traffic being deprioritized that would be even more interesting. And what about Cogent’s wholesale customers. Were they told their traffic was being deprioritized as well?
As I have been saying for many months, the vast majority of people are laying blame on specific companies without having all the data to really make an informed decision. This new admission by Cogent is just another example of companies playing games, in an effort not to disclose everything that is really going on behind the public’s eye. As far as I am concerned, all companies share the blame in this as they all have to work together to fix the problem and keep consumers happy. But for all the lawyers and policy makers that jump on something the moment it is released, and do a blog post laying blame, I’d like them to now explain what is going on when Cogent has now confirmed they put Netflix traffic in the slow lane.