Traffic Management and the Common Framework

We’ve been writing a lot about Internet traffic management here, and lo and behold the common framework for Internet regulation unveiled today by Google and Verizon deals with the subject in fairly specific terms. The proposal says:

Non-Discrimination Requirement: In providing broadband Internet access service, a provider would be prohibited from engaging in undue discrimination against any lawful Internet content, application, or service in a manner that causes meaningful harm to competition or to users. Prioritization of Internet traffic would be presumed inconsistent with the non-discrimination standard, but the presumption could be rebutted.

This is a very dense statement that needs to be parsed carefully as every word is important. It’s a legalese statement, so the words don’t mean the same thing that they do in common language as well. Some of the most important concepts are: “undue discrimination,” “lawful,” “meaningful harm,” “presumed inconsistent,” and “rebutted.”

Google has its own Content Delivery Network of a million or so CPUs around the world that can stream YouTube and other high-bandwidth data streams across short paths to all the major ISPs in the world. This network set Google back some $5 billion, and if you want to compete with YouTube, you need something similar. A startup in the video streaming business – and there aren’t many aside from Netflix – would need to buy CDN services from somebody like Akamai in order to compete. And that’s what Netflix does today.

Now what do you need to do if you want to start a video conferencing service that specializes in high resolution, multi-party conferences? I’m not talking about Skype, which has a nice system for low-res multi-party in beta today, but something that creates the kind of experience you get from Cisco’s Telepresence, so realistic that you feel like you’re in the same room as the folks you’re conferencing? You would need to build a network of very high capacity pipes, with outlets close to all the major ISPs that provided service to your users. It wouldn’t be possible for you to use a CDN, because conferencing by its nature does not deal with storable, cachable content; a video conference is not a web site, and there’s no content to store. That’s quite an investment, so instead of buying your own wires, you’d most likely begin with bandwidth leased from transit providers under Service Level Agreements that provided you with bounds in latency and jitter. And you would essentially be re-selling low latency, low jitter transport as part of your service to your end users. That’s a perfectly legitimate offering on the Internet today, and it would continue to be legitimate under this framework, as it should.

This is an example of the fact that all services are not equal on the Internet, but it comes from the realm of the Internet core, not from the edge services that the net neutrality people want to regulate. If this sort of service is legitimate in the core, it’s reasonable to ask why it shouldn’t be legitimate at the edge. The answer that you get from pro-regulationists, including people at the FCC, is that the market for core services is fully competitive while the market for edge services, especially for consumers as opposed to businesses, is less competitive. They also claim that each ISP has what amounts to a monopoly over access to each consumer, so they need to be more heavily scrutinized than core services providers.

In practice, this means that consumers have a more constrained set of choices in terms of their service plans than businesses do, although most consumers are free to purchase “business class” services if they want, even from the cable companies like Comcast and Time Warner Cable. So the section on “pay for play” high priority services is fairly muddled because this whole priority thing is foreign to the career telecom regulators who have crafted the basic rules for net neutrality.

Picking up our video conferencing example, let’s say after building your massive video network, you find that some of your customers are unhappy with the service, you investigate, and you find that some ISPs don’t have enough capacity to carry your packets properly inside their networks. What do you do about that? One approach that seems fairly straightforward is to offer the ISP money to boost the priority of your packets over the generic packets they’re moving today, mainly to and from web sites, YouTube, and Skype. The money the ISP collects – a portion of your profits, actually – would enable them to upgrade their network long term, or to put the segments your customers are on at the head of the upgrade queue. Should that be legitimate?

Engineering-wise, it’s A-OK, but as a regulatory matter, you’ve got to deal with the FCC to convince them – rebut the presumption – that this form of paid prioritization is non-discriminatory. I can more easily convince a packet-switching engineer of that than I can a career telephone network regulator.

So that’s one of the issues that will need to be resolved before this proposal, or something like it, becomes law.