Ou: Google hypocritical on neutrality

George Ou writes that quality-of-service distinctions, far from being the bane of the Internet, are a Good Thing.

Edited to deal with context problems.

George Ou, who has long railed argued against net neutrality schemes that ban QoS and tiered access and in favor of "reasonable network management," offers a different perspective on whether Google's content caching deals are hypocritical, as a Wall Street Journal article argued (reported here. Google has always been hypocritical, George writes.

The point is that Google is pushing hard for the Markey-2006 and Snowe/Dorgan 2006 proposals that would ban QoS prioritization based on the source (applies to the content provider or the broadband customer since the legislation doesn’t specify). Those bills would also prohibit surcharges on QoS prioritization and Google was sure to leave an exception for all other Internet related businesses by specifically targeting broadband in the legislation. This ensures a exemption to content caching (usually in the form of Content Delivery Network (CDN) providers like Akamai or LimeLight) which offers not just a “fast lane”, but a warp speed lane that operates at an instantaneous speed because they bypass the need to retransmit data. Content caching operates 10,000 times faster (the typical number of clients each caching server services) than the type of QoS prioritization Google is lobbying hard to ban.

Opponents of Net Neutrality regulation have consistently argued that neither type of prioritization is bad and that they’re needed to optimize the Internet, but Net Neutrality regulation would ban QoS prioritization which is critical for making networks better at multitasking and critical for eliminating real-time application killing jitter (I explain this in my new Network Management paper released last week). Google and all their Net Neutrality proponents like Larry Lessig said that QoS prioritization would create a for-fee fast lane that would make it impossible for smaller content producers to compete on an equal scale. But that’s nonsense because QoS prioritization is useless for large-scale content delivery because it does not have the infinite-speed congestion-bypassing characteristics of content caching. A ban on QoS prioritization would simply prevent broadband networks from doing reasonable network management that enhances the value of broadband networks. The type of prioritization technology that does give larger content companies like Google the ultimate advantage in content delivery is conveniently ignored by the proposed Net Neutrality regulations which isn’t surprising when we consider who is backing the legislation.

The key point is that if Net Neutrality proponents want to prohibit companies from gaining a content distribution advantage through financial might, then they should have a problem with content caching and not QoS prioritization. If they truly believe in their cause to create a state of equality that never existed on the Internet, then they should be calling for a ban on caching technologies. But that would be just as silly as banning QoS prioritization.

The reality is that the Internet has always been an open platform for any one, any use, and any business model, but participation has always required varying levels of payment for varying levels of service. BOTH form of prioritization technologies are crucial to the Internet and it would be foolish to ban either technology.

This is precisely why I've released a reportdebunking these myths about network management and QoS technology. Network management with QoS prioritization technology has 3 goals.

  1. It means equitable bandwidth for customers of the same service tier. Equitable bandwidth doesn't mean equal bandwidth at any instantaneous point in time; it means we try to ensure that everyone gets equal average bandwidth over an interval of time such as 15 minutes.
  2. QoS makes networks better at multitasking and support multiple applications with the best performance possible for all users and applications. That means real-time low-bandwidth applications get higher priority than interactive high bandwidth applications like web browsing. Then low duration interactive applications like web browsing should get higher priority over background applications with high duration. That is not discrimination towards background applications like Peer-to-peer (P2P) because the the P2P application still gets higher average bandwidth from the network without the toxicity to other applications.
  3. QoS fixes the jitter problem which can occur at very low network utilization levels. This is where some applications will burst on the network and monopolize the packet queue and starve all other applications for hundred of milliseconds which disrupts real-time applications like VoIP and online gaming.

George also wants to clarify exactly what his position is. Here you go:

I’ve long argued against Net Neutrality proposals that ban tiered services and argued for reasonable network management. This position is now shared by Professor Larry Lessig, a long time Net Neutrality advocate, who now says he opposes Net Neutrality legislation such as the Markey bill which bans charges on QoS. Tim Berners-Lee also argued that paying more for better QoS should be legal. So my position really isn’t all that different from leading Net Neutrality proponents. For example, I’ve argued for a rational debate on Net Neutrality, which was well received by moderates on both sides of the argument.