Peer pressure

Streaming Video Netflix has been doing its best to stay out of the middle of the smackdown between Level 3 and Comcast, having judged — correctly in my view — that this particular dispute is not their fight. Notwithstanding Level 3’s efforts to portray Netflix as the victim anti-competitive behavior by Comcast, unless Comcast has some sort of death wish for its merger with NBC, which is still pending before federal anti-trust regulators, it’s hard to make the case that Comcast was targeting Netflix content per se in demanding higher payments from Level 3 (although video streaming expert Dan Rayburn has done his best to make precisely that case). The conflict still seems to me fundamentally a commercial dispute between two Internet infrastructure providers over how, how much and at what price they will pass bits back and forth between their respective networks–a conflict exacerbated but not caused by the surge in online video (those interested in the pro-Comcast perspective will want to read Stef van der Ziel’s post about the dispute on his CDN Strategy blog).

Just because Netflix was not the intended victim, however, does not mean it has no stake in the outcome. In fact, all online video providers are likely to find themselves drawn ever-deeper into the murky world of transit and peering agreements between Internet backbone and last-mile operators as the amount of video consumed over IP networks increases exponentially.

And I do mean murky. The Internet, by definition and design, contains an effectively infinite number of points where two or more privately owned networks meet and pass bits back and forth. Apart from the TCP/IP technical protocol however, the precise rules of engagement have never really been codified, at least formally. There are no government regulations in the U.S. or elsewhere that govern such exchanges and no industry body exists to promulgate universally accepted “best practices.”

Instead, the great, worldwide Internet has largely gotten by up to now on the same sort of basic code of fairness that prevailed among the engineers who designed and built it. Sending bits over a network, or even between networks, by itself does not cost much. But the infrastructure required to move them represents a significant capital investment by the network operator. Data exchanges between networks, therefore, can have significant cost implications for the respective network operators depending on how much traffic is involved.

Generally speaking, and in grossly simplified terms, operators typically handle such exchanges in one of two ways. Where two networks are exchanging roughly equal amounts of data to and from each another, the operators will typically agree to waive any financial settlements between them on the theory that each is imposing roughly the same amount of cost on the other, and that the costs of administering payments back and forth would not be worth incurring. Where the traffic loads are out of balance, however, and one network is sending substantially more data to another than it is getting back, the operator with the higher volume typically makes payments to the other.

Notwithstanding those rough-and-ready standards, disputes between operators arise all the time. Typically, these end up being handled in the background, out of view of end-users or application developers (to say nothing of regulators) because in the great scheme of the Internet they generally amount to small beans and it usually isn’t in the operators’ interests to drag their customers into it.

The fact that the Level 3-Comcast dispute has burst into the open, however, suggests tensions between network operators are getting harder to contain, and that the old gentlemen’s agreements around peering may no longer be adequate to the task. And online video is a major factor causing those tensions to rise.

The surge in traffic loads network operators are experiencing today, driven largely by the increase in online video use, is putting new strain on operators’ capital budgets and return on investment.

Cisco estimates that total Internet traffic will triple by 2014, to 64 Exabytes per month, and that 90 percent of that volume will consist of video. A recent study by Juniper Networks predicts that Comcast and other last-mile carriers will see traffic increase by 27 percent annually between 2010 and 2014, and that the carriers will need to increase their investment by 20 percent per year to keep pace.

At the same time, however, Juniper predicts that broadband revenue for carriers will increase only 5 percent per year over that same period. According to Juniper’s math, the carrier’s basic business model will collapse in 2014, when annual investments exceed annual revenues.

Examining the same problem from the perspective of wireless carriers, Sanford Berntstein analyst Craig Moffett concluded that revenue per megabit will fall from 43 cents today to 2 cents by 2014.

Coming as they do from a makers of network equipment, the Juniper and Cisco forecasts have to be taken with a grain or two of salt. But it’s no big stretch to say the trend lines are ominous for operators. The problem they face is not fundamentally a question of capacity; it’s a question of economics. As Stuart Elby, vice-president for network architecture at Verizon told Businessweek, “I can find ways to build the capacity, but it’s hard to justify it if I’m not reaping any benefits in terms of revenue.”

Fortunately for operators, FCC chairman Julius Genachowski threw them two possible lifelines last week when he outlined his proposed net neutrality plan, which the commission is scheduled to vote on Dec. 21. Each of those lifelines hold potentially significant implications for online video providers, however.

First, Genachowski would have the FCC bless usage-based pricing (UBP) for broadband service, a common model in Europe but largely unknown in the U.S. among wireline broadband providers. That could increase carriers’ revenue and shift more of the increased infrastructure cost to those users most responsible for driving it. But it could also seriously curtail online video usage, at least initially. Moffett’s analysis showed that, among wireless subscribers, UBP would cause them to reduce their use of virtually all applications, but that wireless video would be the most heavily impacted application.

Subscription service providers like Netflix and Hulu Plus could be particularly vulnerable because UBP would fundamentally change their flat-price, all-you-can-eat value proposition to consumers. Suddenly, the $7.99 per month subscription price would be just the starting point for figuring out how much such a service would cost you to use.

Genachowski also left the door open for network operators to offer so-called  managed services, so that latency-sensitive applications like voice and streaming video could be given priority over less sensitive applications like email and web surfing, presumably for a price. Again, that could let operators shift more of the increased infrastructure cost to those applications most responsible for the increased traffic: video. But it could also lead to increased costs for online video providers.

More subtly, but perhaps more critically, it could put online video providers smack in the middle of peering negotiations. One question hanging over the feasibility of managed services has always been whether their integrity can be maintained from end to end. It’s one thing for Netflix to secure priority service from its first-mile Internet access provider to guarantee quality-of-service (QoS) for its subscribers. But it’s another thing entirely to ensure the integrity of that QoS agreement once the bits pass, as ultimately they must, from Netflix’s first-mile provider (or CDN) to another network operator.

Apart from not explicitly banning the sale of managed services, nothing in Genachowski’s proposed net neutrality framework would appear to address the question of whether the managed services sold by one network operator must be honored by another. Instead, it appears as if the integrity of managed services will still be subject to the sort of (increasingly contentious) negotiations over peering that recently burst to the surface with Level 3 and Comcast.

Online video providers, then, could find themselves facing a situation in which network access providers are able to charge them a premium to guarantee QoS, while other access providers charge consumers higher fees for accessing the content, and neither end of the pipeline has any obligation to honor the commitments of the other. Except, of course, for a price, to be paid either by the video provider or the consumer. Or both.