A fear-monger’s view of networking

Public Knowledge is a backer of the wacky “Save the Internet” campaign, and the following commentary by their leader Gigi Sohn is eye-opening. Sohn comments on the AP story below about Internet congestion from HDTV, totally distorting the facts:

I’m sorry to say that this latest rationale for discrimination doesn’t wash either. These fears (how ironic that the pro-NN forces are the ones accused of fear-mongering) are based on absurd assumptions about how people use the Internet — that people will start watching streaming video like they do regular TV — for 8 hours a day, or that, as the AP story states “everyone in a neighborhood is trying to download the evening news at the same time.” We know that this won’t happen any time soon, if ever. Also, and this should be no surprise, the telcos are not revealing how much it would really cost to provide the best solution to the problem — building a fatter pipe.

There’s no need to assume that people will be watching TV for 8 hours a day for HDTV to represent a significant increase over TV downloads on today’s Internet because the baseline is virtually zero. How much TV does anybody download today? And yes, even one hour of HDTV consumes more bandwidth than the the typical Internet user consumes in a week, and a single baseball game in HDTV is a million times larger than a web page. As for people downloading news at the same time, isn’t this exactly what will happen if the broadcasters and telcos don’t go to Internet multicast in real-time for broadcasting? The news doesn’t have much value hours later.

Indeed, Internet2’s Gary Bachula testified a few months ago that the best, and most cost-effective way to deal with any capacity issues is to make the pipe fatter. Internet2, which is the next-generation Internet available only at Universities and colleges, considered both a discrimination-based model and a 100 MG pipe model, and chose the latter. If telcos and cable companies are permitted to create a two-lane Internet w. tolls for access to the high-speed lane, they will have no incentive to build the fat pipe, because they would then lose the revenue from the high-speed lane.

Bachula’s testimony in probably the most misunderstood comment on QoS in recent memory. His experimental network doesn’t serve a consumer market. It’s built on an over-provisioned backbone and fat pipes to the offices of the professors and grad students who connect to it. His discovery is nothing new: an over-provisioned network doesn’t take much management. But out here in the real world were people pay for services and companies have to profit from providing them, we don’t have such luxury. Solving QoS by adding bandwidth is like improving your computer’s performance by adding memory, it only works until the next release of Windows and then you need more memory again. QoS is needed on large scale networks that operate in the normal and economic range of bandwidth.

In the short term, there are other non-discriminatory ways to deal with this so-called “choking” of the Internet. One way is to put a cap on he amount of data that a user gets for free, and then charge extra if that person uses more, like cellphone usage.

There really aren’t any “non-discriminatory ways” of managing a network, as any user who finds what he wants to do constrained by policy is going to feel victimized.