AT&T (s t) recently announced the elimination of one of broadband Internet’s most prized features: unlimited use at a flat rate. While the trend toward metered bandwidth is not inherently pro-consumer, ISPs have staked out a singular public rationale: data caps are necessary to limit the consumption of “bandwidth hogs” in order to protect the network experience for everyone else. Such concepts are simplistic and easy to imagine. They are also completely wrong.
In the fixed-cost network model (used by most ISPs here in the U.S.), there’s very little connection between raw consumption levels and the relative cost of serving consumers. The heaviest of users may often be the most profitable customers, depending on when they consume network resources.
Wait. What? Heavy users are the most profitable?
Yes. Because overall congestion, not individual consumption, is the single driver of network costs. It’s not the “how much” but the “when” that really matters.
A fixed-cost network is analogous to a highway system. Highways must be designed to handle peak traffic, which in most cities is rush hour.
This establishes the initial cost of building out the highway network. When traffic grows, new lanes have to be built and new costs are added to the equation. It is not the 2 percent of cars using the empty roads at 4 a.m. that creates the demand for new lanes.
It’s the same for your ISP’s network. On your ISP’s network, “rush hour” is called “peak time,” with congestion usually occurring between 7 p.m. and 10 p.m. The capacity required to handle these peak times sets the cost benchmark.To continue with the highway analogy, would rationing gas or capping miles on everyone’s odometer solve rush hour traffic?
No, of course not. People would still prioritize use of their car and precious miles to get to and from work. They might take fewer road trips on the weekends when the roads are already clear. Rush hour would still exist, and the road would be no cheaper to build or expand. Likewise, bandwidth caps won’t relieve congestion at peak times.
No one wants a global Internet riddled with traffic jams. Not users and certainly not video-on-demand services, cloud companies, MMO games, or any other services requiring bandwidth. The fact that these services require more bandwidth has been a boon for ISPs. They drove consumer demand for broadband, which is much more lucrative than dial-up Internet. Therefore, we all have a stake in ensuring a healthy Internet.
But if not data caps, then what?
BitTorrent is in a unique position to understand network challenges and usage patterns. It has over 100 million active monthly users in every connected country in the world. This gives us significant insight into network speeds and congestion. Our data indicates the solution is twofold: High-bandwidth services need to be good citizens and peak usage needs to be brought under control.
Yield to Priority Traffic
Since any data traffic that doesn’t induce congestion on a fixed cost network is essentially free; applications can voluntarily play a role in traffic prioritization. And since BitTorrent is a high percentage of global Internet traffic, we have a responsibility to be a part of the solution.
This was the primary motivator around our release of a new protocol a year ago, called µTP. The protocol essentially senses congestion and self-regulates to avoid contributing to Internet traffic jams.
Because µTP can never induce network congestion, it doesn’t contribute to an ISP’s cost. An ISP still has regular network maintenance expenses, but remember, with a fixed-cost network, traffic only becomes an economic burden if it contributes to congestion and forces the need for expansion.
As a result, µTP is exceedingly friendly to ISPs and their business model. µTP is open-source, and we invite application and cloud services providers to work with us directly or in the IETF’s LEDBAT working group in the ongoing innovation and usage.
Voluntary good citizenry on behalf of the industry is a start, but it’s not a complete solution to congestion.
One idea is to pursue congestion-based/peak hour pricing. Like the fee to enter London streets during rush hour, congestion-based pricing is a reasonable compromise to alleviate Internet logjams at the times it’s needed most, while letting consumers surf freely the rest of the time. For example, your Netflix (s nflx) movie rental could be set to download overnight if you plan ahead. And if you don’t? Perhaps you pay a premium to download it on-demand on a Saturday evening.
While congestion-based pricing directly targets the incentive structure around the problem, we recognize it would be difficult to implement. The CONEX working group in the IETF — focused on developing a mechanism by which senders inform the network about the congestion encountered — is aimed in this direction but implies wholesale changes to the current Internet economic models. At BitTorrent, we believe any solution targeting the actual problem is worthy of serious consideration.
Ultimately, this isn’t just an ISP challenge.
Keeping the Internet healthy for consumers and a place for prosperous innovation of new services must be the top priority for all of us – ISPs and the technology industry alike. Limitations will stifle the growth and potential of the Internet. Instead, let’s look at real-world technical and business solutions to serve the needs of our mutual customers and protect the infrastructure on which we all depend.
Eric Klinker is the CEO of BitTorrent, a file-sharing application.