University of Minnesota
November 6, 2009
Andrew Odlyzko keeps tabs on Internet traffic.
Observations on the growth of online traffic and Internet capacity
By Deane Morrison
With Internet traffic growing steadily, rumors that overusage will crash this pillar of modern society crop up from time to time. But Andrew Odlyzko doesn't buy them.
As principal investigator for the Minnesota Internet Traffic Studies project, the University of Minnesota mathematics professor monitors Internet usage, taking advantage of data from a variety of sources. He sees traffic growing at somewhere around 50 percent per year, but it's nothing to worry about—at least not yet.
"The Internet is growing rapidly, but at rates that can be accommodated by the current rate of improvement in technology," he says. In fact, the rate of growth even seems to be diminishing. In particular, he says fears of "exafloods" of traffic, brought on largely by video streaming, are unlikely.
Such worries have caused some Internet-related businesses to consider, or advocate, a new pricing system based on use. If that happened, instead of paying a flat monthly fee, users would pay "by the byte." (Odlyzko estimates that average consumers use between 10 and 15 Gigabytes per month; high-end users can go up to 250 Gigabytes per month.)
A constant conflict exists between efficiency and fairness in markets, he says, and the Internet is no exception. The fairness issue is embodied in the "net neutrality" concept, which means no one owns the Internet and carriers can't force users to, say, switch from their favorite search engine to another one. But, says Odlyzko, net neutrality doesn't preclude charging according to use, and that's where the potential for conflict comes in.
But, says Odlyzko, "if you charge by the byte, usage drops. Metered pricing is a way for industry to shoot themselves in the foot." He also can count numerous historical instances where the public protested loudly against what it perceived as unfair pricing schemes.
The unlikelihood of exafloods due to video streaming and similar practices is based on data showing that they are not straining the system.
"It's true that if public decided to abandon TV, or satellite TV, and get all their movies and news stories through the Internet, it would crash," Odlyzko says. "However, we observe that people are not all that fast to change their habits. For example, about a third of the population has digital video recording, but even these people mostly watch traditional TV or cable."
From the industry standpoint, the real concern is people who upload material, especially pirated material, which then gets copied and spread around, he says. Those practices can cause traffic to mount very quickly.
For now, then, no crisis is looming for the wired Internet. "There's no need for a change in architecture or pricing," says Odlyzko. But wireless is a different kettle of fish.
"The explosion in data usage on wireless networks, such as from smart phones and BlackBerries, is putting a strain on the networks," he says, and it could lead to more constraints on the plans offered to consumers.
What's needed—and what Odlyzko does—is to collect information that reveals the balance between supply and demand so decision-makers can do the right thing, he says. He cites the examples of e-mail and texting; the popularity of both was predictable, but took the industry by surprise.
"You need to collect data because people sometimes do unexpected things," he observes.