http://arstechnica.com/tech-policy/news/2011/12/data-caps-a-crude-and-unfair-tool-for-easing-online-congestion.ars Internet providers argue that they need to impose monthly data caps on their users in order to slay the "bandwidth hogs" running wild and free through their networks, goring ordinary users with their tusks when all those users want to do is view some funny cat pictures online after a tough day at the office. The idea is that a monthly quota can reduce the amount of network congestion during peak hours throughout the month. Fact or fiction?
One piece of new research argues that it's fiction. "Our analysis confirms that data consumption is at best a poor proxy for bandwidth usage," writes Benoît Felten, chief research officer of Diffraction Analysis.
Two years ago, Felten and Herman Wagter (the man who spearheaded Amsterdam's fiber rollout and wrote about it for Ars) issued a challenge to Internet providers: show us the evidence.
"Any telco willing to actually understand what's happening there and to answer the question on the existence of hogs once and for all can extract that data and send it over to me, I will analyse it for free, on my spare time," Felten wrote in 2009. "All I ask is that they let me publish the results of said research (even though their names need not be mentioned if they don't wish it to be). Of course, if I find myself to be wrong and if indeed I manage to identify users that systematically degrade the experience for other users, I will say so publicly. If, as I suspect, there are no such users, I will also say so publicly. The data will back either of these assertions."
A midsized American DSL provider finally stepped up to the plate, offering fine-grained detail over the course of a single day from one aggregation link that served 5,138 users. Felten and Wagter broke down the daily data consumption into five-minute increments and went to work.
You're doing it wrong
Their detailed analysis is available as a paid report, but Felten did make his conclusions public this week. He found that 48 percent of active Internet customers "are amongst the top 10 percent of bandwidth users at one point or another during peak hours." Controlling real-time congestion by going after just a few high-data monthly users, then, is unlikely to be effective.
"Data caps, therefore, are a very crude and unfair tool when it comes to targeting potentially disruptive users," writes Felten. "The correlation between real-time bandwidth usage and data downloaded over time is weak, and the net cast by data caps captures users that cannot possibly be responsible for congestion."
His solution: look only at those causing actual congestion during periods of peak use—generally four to five hours in the evening. (Comcast has employed such a system, though it also uses data caps.)
What Internet users do outside that time should have little effect on other network users (because the aggregation links aren't even close to saturation) or on the Internet provider (because the marginal cost of additional traffic on one's own network is essentially zero, and peering and transit arrangements with other ISPs and backbone providers generally involve paying for bandwidth rather than data). So data caps serve as a general warning to subscribers against excessive use—whatever that is—while doing very little to address actual congestion problems.
It's noticeable that ISPs rarely show evidence to back up their claims regarding bandwidth control by throttling being effective or what attempts they are making to improve capacity.
