Drowning in the Exaflood

A colleague sent me this YouTube video:

Although I think it make some good points, I think it gets a couple of things wrong. One is the point about “smart” networks. I’ve discussed that in detail before, so I won’t belabor it here. The other thing that seemed off to me is this notion that we’ve got an “exaflood” on its way, and if we don’t have enough bandwidth, the Internet’s architecture won’t be able to “keep up” with the demand.

This reminds me of the silly “peak oil” activists who seem to regard oil consumption as inexorably increasing, rather than subject to the forces of supply and demand. The consumption of oil is sensitive to its price, so if oil gets more scarce, the price will rise and people will economize. It doesn’t make sense to talk about “running out” of oil.

In an analogous way, network applications are deployed when the infrastructure exists to support them. If you had launched YouTube in 1995, it wouldn’t have worked, both because the server bandwidth would have been too expensive, and because most consumer wouldn’t have had enough bandwidth to receive the videos in a reasonable manner. As bandwidth becomes more plentiful, its price drops and applications that use more of it become economically feasible. The “exaflood” will happen when and only when the Internet’s infrastructure has been upgraded enough to handle an exabit data rate. The demand for data is not an independent variable that network owners have to “keep up” with.

The weird thing about this is that I think there’s a much more natural way to make the same point, which is simply that we want to ensure telecom companies have adequate incentives to upgrade their networks. Not to deal with the looming “exaflood,” but simply because faster networks allow us to do more and better things with our Internet connections. The problem with framing it that way, I guess, is that it doesn’t create a sense of crisis, and it’s hard to get people excited without a crisis. Also, I suppose, it accentuates that the “exaflood” isn’t a new problem, but simply a continuation of the same steady expansion of the Internet that’s been under way for more than a decade now.

SHARE: