From the Core to the Edge: Perspective on Internet Prioritization

The House Communications and Technology held a great hearing on Internet optimization today. The hearing addressed the core issue in the net neutrality controversy, the ability of ISPs to optimize traffic streams for a fee.

This practice – which is more theoretical than real – has been made into fear fodder by the economic interests behind net neutrality as well as their non-profit enablers. Witness Paul Schroeder of Aira Technology presented his company’s app as a use case for managed wireless traffic.

This app provides visually impaired people with a guide who can describe their surroundings to them. Aira tried doing this with ordinary wireless and found that it didn’t work, but they now use a service from AT&T that brings it to life.

My Testimony

I submitted 40 pages of testimony to the committee that comprises a tutorial on Internet optimization and its uses. My first draft had a factual error – I said Tom Wheeler created the paid prioritization bugaboo in 2015 – and the committee was gracious enough to allow me to correct it.

Here’s a summary of my five minute opening remarks as prepared. I didn’t exactly stick to the script because I don’t know how to do that.

  1. Prioritization has been part of the Internet’s design from the beginning, but is hasn’t always been part of its practice. Type of Service, IntServ, and Diffserv are examples. It was controversial for a time, but those days are behind us.
  2. After 15 years of discussion, we’ve reached consensus on the fact that it’s legitimate for ISPs, CDNs, transit networks, and purpose-built services like Webex to accelerate time-sensitive Internet apps such as enterprise voice and video conferencing.
  3. We also appear to appreciate the power of competition to build cost-effective networks that optimize delay, throughput and reliability for both content-oriented apps and real-time apps.
  4. We realize, I believe, that as Internet routers have become more powerful, the reach of the Internet has grown, and the pool of Internet applications has expanded, prioritization and related Internet optimization techniques such as resource reservation, traffic shaping, and dynamic path selection have become not only commonplace but essential.
  5. This is good because no matter how much capacity networks have, there will always be opportunities to improve. Many are waiting with bated breath for 5G, which promises so much. But I expect 6G will be even more awesome. Technology always marches forward.
  6. I think we appreciate that prioritization mechanisms such as the IEEE 802.11e standard (which I helped design) are beneficial to real-time applications such as voice in the same way that LTE bearers are. The fact that one is provided “free” on closed enterprise networks while the other is sold to all interested parties is irrelevant to their utility.
  7. The remaining disagreements about prioritization and other Internet optimization techniques seem merely to be questions about price and regulatory consistency. I believe it goes without saying that all firms who provide a given service should generally be regulated in the same way.
  8. It’s hard to explain the continual increases in broadband speeds we’ve seen in the US over the last ten years – speed improves 35% per year – without giving some credit to the expectation of profit. The fact that web speeds have stagnated over this period – even declining in 2016 – suggests something is wrong with the web’s financial model.
  9. Leaving the consumer broadband market aside, Internet optimization is important to enterprises that have to connect branch offices to data centers located at corporate headquarters or in the cloud. These systems support telephone calls, video conferencing, software as a service, and access to corporate databases. Traditionally, branch offices have relied on private lines, but public Internet connections are much less costly.
  10. Instead of paying $300/month for a 1.5 Mbps T1 connection to HQ, branch offices can get 50 – 250 Mbps connections for less than $75, with the flexibility to access the Internet from them as well. But they require management to ensure high quality video conferencing and enterprise voice, however. The option to purchase a managed service that prioritizes access to headquarters over Internet access on the same wire hasn’t always been clear; it would be good for Congress to make it so.
  11. If firms cannot specify Service Level Agreements for data services that combine Internet use with private Business Data Services on the same wire, their networking costs will be artificially high and innovation will suffer. Business data is a more competitive market than consumer internet is in the pre-5G era; this market can probably police itself.
  12. The Internet is not simply a sandbox for network research any more, it has become the primary means of electronic communication around the world. Before long, it will be the only such means and we will all be better for it. Please allow firms that depend on networking to invest efficiently so as to maximize their incentives to innovate.

We all gain from advances in internet technology. Thank you, and I look forward to your questions.

The Video

You can see the hearing on YouTube. It’s a shame we didn’t do this years ago; please bear it in mind next time you hear someone describing net neutrality as an “all packets are equal” rule.