• by Wendy Davis  @wendyndavis, September 22, 2008

Comcast confirmed on Friday that it is testing a new traffic management system that involves slowing down the connections of some subscribers who use a disproportionately high amount of bandwidth.

The new approach replaces Comcast's prior method, which involved slowing peer-to-peer traffic. The Federal Communications Commission, led by Chairman Kevin J. Martin, ruled 3-2 last month that Comcast's decision to target specific protocols violated the principle that Internet service providers should treat all lawful applications equally.

Comcast has appealed that ruling, but the company says it intends to change its traffic-shaping practices. The new system focuses solely on the quantity of bandwidth that individual users consume, and not the type of application.

"Importantly, the new approach will be protocol-agnostic," Comcast wrote in papers filed Friday with the FCC. "The new approach will focus on managing the traffic of those individuals who are using the most bandwidth at times when network congestion threatens to degrade subscribers' broadband experience."

As part of its new traffic-shaping system, Comcast will create two classes of traffic--a default category dubbed "priority best effort," and the potentially slower "best effort" category. Comcast will deprioritize subscribers who use a high amount of bandwidth in a 15-minute period, placing those people in the "best effort" group. If they decrease their bandwidth use for a 15-minute period, they will be returned to the "priority best effort" state.

The company said that subscribers who have been downgraded to "best effort" might not notice any change if congestion overall isn't high. But during times of high congestion, they might see a difference. "Typically, a user whose traffic is in a (best effort) state during actual congestion may find that a Web page loads sluggishly, a peer-to-peer upload takes somewhat longer to complete, or a VoIP call sounds choppy," the company wrote.

Comcast said it has tested the new methods in five markets--Chambersburg, Pa.; Warrenton, Va.; Lake City, Fla.; East Orange, Fla.; and Colorado Springs, Colo.--and has not received any complaints.

The total number of people affected so far, however, appears small. Comcast said the test in Colorado Springs involved only 6,016 subscribers. Of those, an average of just 22 subscribers a day were deprioritized. In all five test markets, around one-third of 1% of subscribers were affected.

Comcast's filing was in response to an FCC order of Aug. 18 directing the company to outline a new traffic management plan by midnight Friday.

The FCC began investigating Comcast after receiving complaints from companies and advocacy groups, including Free Press and Public Knowledge.

Chairman Martin, a Republican, proved to be one of Comcast's most vocal critics on the issue. He condemned the company at a public hearing in February for instituting traffic management practices without disclosing them, and later joined with the two Democrats on the commission to rule that Comcast had violated net neutrality principles by targeting peer-to-peer applications.