Another Study Shows Data Caps Are Likely Ineffective, Address Wrong Problem

Data caps on your broadband, while in principle sound troublesome, are at least understandable. Bandwidth is a limited resource and we all have to share it, and presumably if we all were maxing our connections out all the time, we’d tax the system beyond its capacity. But who uses the most bandwidth and when is a more practical thing to investigate, as knowing that could prevent congestion at peak hours and so on.

Some studies and theories have suggested that so-called bandwidth or data hogs, in other words people who use the entirety of the product they paid for, aren’t really a great source of congestion, and the data caps intended to prevent such users from maxing out all the time aren’t an effective countermeasure.

The guys at Diffraction Analysis examined data from “a mid-size company from North America” that was interested in understanding its consumers’ use patterns. Good for them, by the way. The data they submitted was bandwidth consumption throughout the day, with five-minute granularity. The study’s aim was to determine whether a small subset of users (the hogs) could indeed affect the quality of others’ service, and whether caps were an effective deterrent.

The conclusions, briefly stated, were that while heavy users do in fact consume far more data in aggregate than the average (288GB vs. 9.6GB in this study), their contribution to congestion during peak hours, and when the network is at 75% of its capacity or above, is in fact not much greater than the average user.

What the statistics bear out is this: during peak hours when service is most likely to be affected by overcrowding, heavy users only make up a small percentage of those consuming bandwidth – 14.3%, to be precise. And of the heavy users, only half of them were on the fastest connection, further driving home the fact that while they may consume more in total, they are not contributing more than anyone else to the actual problem, which is slowdown in peak hours.

So why the data caps? Clearly a limit of, say, 300GB a month (or lower) won’t prevent peak usage from affecting service quality. In fact, if people are limited by draconian data caps, they are likely to limit their usage to peak hours: streaming a movie in the evening, or browsing YouTube when they get home from work. This would in fact contribute even more to the problem of peak crowding.

What’s the solution? Bandwidth caps seem more important, and advertising a range of values instead of a maximum would be both more honest and indemnify the ISP against slowdowns. If a dynamic bandwidth cap let you download at 30Mbps in the middle of the night but limited you to 5Mbps during peak hours, it’s the best of both worlds and nobody has to worry about overage charges.

And how would you make money to replace those overages, not that they amount to much? Sell a limited number of premium accounts that aren’t limited during peak hours. Since the ISPs control the number and width of the pipes, they can calculate how many premium and how many standard they can offer. This seems much more logical than imposing a total data limit that’s a pain for some and immaterial to others, though both contribute equally to the problem ostensibly being addressed.

The whole report is available for purchase here for the sum of €750, though the executive summary provided by the author is illuminating as well.