Netflix uses adaptive streaming for its standard quality video and this only requires about 2 Mbps at the customer end to get the quality that Netflix intends for the video. HD videos require more bandwidth, but customers are complaining about standard video. A Netflix download requires a burst of data up front so that the movie can load ahead of the customer. But after that it stays steady at the 2 Mbps rate and the download even pauses when the customer pauses. It’s getting hard to find an urban ISP that doesn’t deliver at least that much speed, so one would assume that any customer who subscribes to at least 2 Mbps data speeds should not to be having trouble watching Netflix.
But they are. On their blog Verizon talks about a customer who has a 75 Mbps product and who was not getting good Netflix quality. On that blog Verizon says that it checked every bit of its own network for possible choke points and found none. For those not familiar with how networks operate, a choke point is any place in a network where the amount of data traffic passing through could be larger than the capacity at the choke point. In most networks there are generally several potential chokepoints between a customer and the outside world. In this blog Verizon swears that there is nothing in its network for this particular customer that would cause the slowdown. They claim that the only thing running slow is Netflix.
This is not to say that there are no overloaded chokepoints anywhere in Verizon networks. It’s a big company and with the growth of demand for data they are bound to have choke points pop up – every network does. But one would think that their fiber FiOS network would have few chokepoints and so it’s fairly easy to believe Verizon in this instance.
Verizon goes on to say that the problem with this Los Angeles customer is either Netflix or the transit providers who are carrying Netflix traffic to Verizon. Verizon is not the only one who thinks it’s the transit interface between the networks. Here is a long article from Peter Sevcik of NetForecast Inc. that shows what happened to the Netflix traffic at numerous carriers both before and after Netflix started peering directly with Comcast. This data shows that traffic got better for everybody else immediately upon the Comcast transition, which certainly indicates that the problem is somewhere in the transit between Netflix and the ISPs.
Verizon says the problem is that Netflix, or the intermediate carriers don’t want to buy enough bandwidth to eliminate chokepoints. Sounds like a reasonable explanation for the troubles, right? But then Dave Schaffer, the CEO of Cogent came forward and pointed the finger back at Verizon. He says that the problem is indeed in the interface between Cogent and Verizon. But Schaffer claims this is Verizon’s fault since they won’t turn up additional ports to relieve the traffic pressure.
So now we are back to square one. The problem is clearly in the interface between Verizon and carriers like Cogent. But they are blaming each other publicly. And none of us outside of this squabble are going to know the truth. Very likely this is a tug-of-war over money, and that would fall in line with complaints made by Level3, who says that Verizon is holding traffic hostage to extract more money from the transit carriers.
The FCC is looking into this and it will be interesting to see what they find. It wouldn’t be surprising if there is a little blame on both sides, which is often the case when network issues devolve into money issues. Carriers don’t always act altruistically and sometimes these kinds of fights almost seem personal at the higher levels of the respective companies. The shame from a network perspective is that a handful of good technicians could solve this problem in a few hours. But in this case even the technicians at Verizon and the transit carriers might not know the truth about the situation.