Comcast Plans Major Shift in Approach to "Network Management"
Comcast, in a joint statement with BitTorrent, announced plans for a significant change in the way it responds to network congestion. As noted in a recent post , revelations last year that Comcast sometimes interferes with its subscribers' P2P upload traffic sparked fears that carriers might play favorites and prompted complaints that led to a pending FCC proceeding on the question of "network management." CDT isn't eager to have the FCC adopt formal rules on network management, but having carriers affirmatively degrade selected traffic poses real risks to innovation and competition. The Comcast/BitTorrent announcement is therefore a welcome development. Comcast now says by the end of the year it will "migrate" to congestion management techniques that are "protocol agnostic." This sounds a lot like what CDT suggested in its comments to the FCC -- namely, that network management practices involving any form of traffic degradation should be evenly applied, rather than singling out specific services or applications. It also sounds like a real departure from the technique Comcast described in its comments, which had Comcast targeting specific P2P applications that had been identified as causing congestion problems in the past. To some extent, the devil is in the details. It's not clear from the announcement precisely what Comcast envisions; the criteria for selecting traffic for degradation and the technical means by which it might be degraded remain open questions. (In CDT's view, there is a pretty simple and obvious answer: target the traffic of users who are using huge amounts of bandwidth, without paying any attention whatsoever to the contents or protocols of that traffic.) Because "protocol agnostic" could be interpreted in several ways, transparency -- another principle CDT stressed in its comments to the FCC -- remains crucial. But here too, the Comcast/BitTorrent announcement is encouraging. It quotes Comcast's CTO as saying that the company will "refine, adjust, and publish" the new network management technique "based upon feedback." In short, initial signs are very promising. But of course, today's announcement is specific to Comcast. CDT would like to see general adoption of the concept that congestion management should be (i) agnostic as to content or protocol and (ii) transparent to Internet users and applications developers. (CDT has also said that network management techniques should be consistent with core internetworking standards.) Endorsement of these principles by other carriers could be an important next step, as could a decision by the FCC to incorporate these principles into its broadband Policy Statement. Finally, it's worth noting a few other positive aspects of the Comcast/BitTorrent announcement. First, it calls for ISPs, technology companies, and the Internet Engineering Task Force to cooperate to develop "a new distribution architecture" for rich media content. I think this translates into coming up with standards for ensuring that P2P protocols are efficient and not disruptive. This too is important, because P2P is rapidly becoming a key distribution mechanism for legal, mainstream, commercial content. Better and more efficient P2P systems may be needed to help minimize congestion problems going forward. Second, the announcement also revealed that Comcast plans to increase upstream capacity for its subscribers, including doubling upstream capacity in some markets by the end of the year. This is good news, because the old assumption that most subscribers are mainly passive downloaders is quickly growing outdated. In a world of user-generated digital content, Internet users are creators too, and the balance between downloads and uploads is growing more symmetric. Third, a small terminology point. The statement refers in the very first sentence to "network capacity management." This is a much better term than just "network management." Much of the policy debate has been confused by the conflation of efforts to fight network threats (viruses, phishing, denial-of-service attacks, spam) and efforts to deal with "bandwidth hogs" who can cause congestion but aren't malicious. The problem with the term "network management" is that it seems to sweep in both. But as CDT noted in its reply comments to the FCC, protecting subscribers from security threats is a very different goal than addressing high bandwidth usage by individual subscribers. The term "network capacity management" puts the focus clearly on the bandwidth hog problem. If it catches on, it could help add some clarity to the discussion.