|
When I first read this post about Predictable Network Solutions on the excellent Telco 2.0 blog, I thought it was an April Fool’s Day hoax. Then I remembered that it’s a UK site, and some Googling confirmed that it’s a real company. So my question is, will this technology—or something like it—eventually make network neutrality a non-issue? Or will it be the means for network operators to implement the discrimination that everyone is worried about?
Basically, Predictable Network Solutions makes technology that dynamically re-allocates network latency and throughput on a per-user basis. It works using statistical traffic models, not application-specific deep packet inspection. In theory, that means network operators will no longer have a network management reason to fear P2P traffic, because it won’t strain the peak capacity of the network. And also in theory, users and application providers will be free to use the network however they please, with better performance on bandwidth-intensive services like video than they experience today.
So is broadband nirvana around the corner? I’m not sure. It’s too early to tell.
Let’s assume the technology works as advertised, and this tiny British startup (or someone else taking a similar approach) succeeds in getting it deployed widely on major broadband networks. Network operators will still have choices in how they manage traffic and pricing. They will still have incentives to bundle their access service with content and applications, or to offer exclusive deals to preferred providers. If you think such arrangements are dangerous, better broadband congestion management won’t prevent them. It would, however, make clear tthat the deals are about revenue opportunities, not traffic engineering. A network operator that wants to exercise tight control over applications and content will see Predictable Network Solutions’ solutions as tools to help achieve that end.
The big lesson here is that technology doesn’t stand still. Network neutrality became a concern partly because application-specific traffic management became feasible. As network management technology develops, it’s likely to reshape the debate yet again.
Sponsored byVerisign
Sponsored byCSC
Sponsored byRadix
Sponsored byVerisign
Sponsored byDNIB.com
Sponsored byIPv4.Global
Sponsored byWhoisXML API
There are actually a number of people working on variations of this technique. The basic idea is to give the end-user the ability to tell the ISP about the latency requirements of his various streams, so that his own BitTorrent streams (or those of other family members) doesn’t interfere with more demanding streams, such as VoIP. So the overall system is like this:
1. User tags streams according to their transit requirements, assigning each to a traffic class.
2. ISP checks volume of packets in each class and demotes or denies any that are in excess of contracted quotas.
3. All packets in the lowest traffic class are accepted, but may be delayed if any transit point is congested.
4. The network run use all provisioned capacity and still provide low delay for the streams that need it.
Neutralists should be happy because it removes the need for ISPs to do DPI, and ISPs should be happy because they don’t have to deploy excess capacity they can never use. Core network providers are happy because they can also sell service levels to ISPs.
This is the way the Internet already works for many commercial users, so the big advance is simply allowing the residential user to have the same set of options.
I gave a little talk about a scheme of this sort in Washington on Mar. 12th, and most people were pretty excited about it. I’ll most likely take it a bit further at Supernova.
My understanding of the relationship between technology and network neutrality is a battle between innovation at the edge and the core (infrastructure) of the network. At the edge innovation, which is the heart of the neutrality argument, is primarily to ensure and enhance the non-distortion of (existing and future) many-to-many Internet applications. Thus, to the proponents of neutrality, it is not a question of innovation (whether technological or otherwise), rather innovations’ potential undesirable impact on the Internet.
The impact of innovation can be better measured by the resulting social welfare rather than, for example, corporate profits. Theoretical models using two-sided markets suggest that innovations at the infrastructure level that, for example, distort pricing decrease welfare.
Hence technology alone cannot solve network neutrality.