At least one piece of the simmering interconnection dispute over Netflix data bottlenecks has been getting clearer lately. Following Netflix's 'crowded network' warning message to subscribers in June, last week Verizon put out a piece demonstrating that its last mile wasn't congested at all. And yesterday, Level 3 turned that around in a blog post, saying that Verizon's explanation is more of an admission.
|Verizon's Chart||Level 3's Chart|
Level 3's prior comments have been more general, not naming Verizon quite so explicitly as doing this -- not that it was much of a secret. Indeed, putting Verizon's and Level 3's charts together shows quite explicitly that the reason for slow Netflix streaming speeds to its customers is a lack of upgrades at their peering connections. It also shows that Verizon's network has plenty of capacity to handle the extra data they would get if they upgraded, and that the lack of upgrade is a business decision not based on the actual costs of that upgrade.
In other words, it's a pressure tactic aimed at forcing payment for traffic. We all knew that of course, and in the world of peering and transit it is hardly a new thing. But perhaps what all this public attention has done is make it so the FCC really can't ignore such things in the context of network neutrality and the leverage held by the last mile. Verizon's remaining position here is simply one of traffic ratios and their role in peering connections, i.e. that anybody sending too much OTT video to them will see traffic ratios rise and trigger a refusal to upgrade peering connections. The only way out will be to pay them directly (in paid peering) or pay someone else who does (a CDN) or they'll use peering upgrade delays to slow it down.
I've long maintained that in a world of streaming video and asymmetric consumer broadband connections, using balanced traffic ratios as a basis for peering makes no mathematical sense. Nor does blaming traffic on one side of a two-party exchange of data transaction for which both parties already pay. Doing so places the industry in a downward trajectory toward a power imbalance in which a last mile monopoly can set its own prices for certain types of data (e.g. video) until they go too far and regulators step in and do it for them.
Yet Netflix *should* be paying for paid peering, and it should want to based on the better performance it would get by cutting out the middle men and all the hops in between. That's why Akamai has done so well for itself in a world where IP transit backbones are so rarely a significant money-making proposition. Netflix's position of free interconnection for all, even those who don't actually have a network to build and maintain, is just as non-sensical. The likes of Level 3 and Cogent and their other customers are just caught in the middle.
Meanwhile, the FCC is getting bombarded with comments on its latest net neutrality foray. Most of them are uninformed and will never be read, but the point is that people really are watching. And the power really does lie with them, which is precisely why Verizon felt so provoked by Netflix's 'crowded network' warning message as to speak publicly about their network traffic for once.
The industry as a whole -- last mile, middle mile, backbone and content -- needs to sit down and work out a new paradigm that works for all. Because if it doesn't and enough of the public feels its internet is under threat, then the eventual solution will be a political one imposed from above. And that really would be a disaster.