If
you’re the customer of a major American internet provider, you might
have been noticing it’s not very reliable lately. If so, there’s a
pretty good chance that a graph like this is the reason:
These graphs
comes from Level 3,
one of the world’s largest providers of “transit,” or long-distance
internet connectivity. The graph on the left shows the level of
congestion between Level 3 and a large American ISP in the Dallas area.
In the middle of the night, the connection is less than half-full and
everything works fine. But during peak hours, the connection is
saturated. That produces the graph on the right, which shows the packet
loss rate. When the loss rate is high, thousands of Dallas-area
consumers are having difficulty using bandwidth-heavy applications like
Netflix, Skype, or YouTube (though to be clear, Level 3 doesn’t say what
specific kind of traffic was being carried over this link).
This isn’t how these graphs are supposed to look. Level 3 swaps
traffic with 51 other large networks, known as peers. For 45 of those
networks, the utilization graph looks more like this:
The graph on the left shows that there is enough capacity to handle
demand even during peak hours. As a result, you get the graph at the
right, which shows no problems with dropped packets.
So what’s going on? Level 3 says the six bandwidth providers with
congested links are all “large Broadband consumer networks with a
dominant or exclusive market share in their local market.” One of them
is in Europe, and the other five are in the United States.
Level 3 says its links to these customers suffer from “congestion
that is permanent, has been in place for well over a year and where our
peer refuses to augment capacity. They are deliberately harming the
service they deliver to their paying customers. They are not allowing us
to fulfill the requests their customers make for content.”
The basic problem is those six broadband providers want Level 3 to
pay them to deliver traffic. Level 3 believes that’s unreasonable. After
all, the ISPs’ own customers have already paid these ISPs to deliver
the traffic to them. And the long-standing norm on the internet is that
endpoint ISPs pay intermediaries, not the other way around. Level 3
notes that “in countries or markets where consumers have multiple
broadband choices (like the UK) there are no congested peers.” In short,
broadband providers that face serious competition don’t engage in this
kind of brinksmanship.
Unfortunately, most parts of the US suffer from a severe lack of
broadband competition. And the leading ISPs in some of these markets
appear to view network congestion not as a technical problem to be
solved so much as an opportunity to gain leverage in negotiations with
other networks.
In February, Netflix agreed to pay Comcast to ensure that its videos would play smoothly for Comcast customers. The company
signed a similar deal with Verizon in April. Netflix signed these deals because its customers had been
experiencing declining speeds
for several months beforehand. Netflix realized it would be at a
competitive disadvantage if it didn’t pay for speedier service. After
its payment to Comcast, Netflix’s customers experienced a
67 percent improvement in their average connection speed.
Netflix has
accused Comcast
of deliberately provoking the crisis by refusing to upgrade its network
to accommodate Netflix traffic, leaving Netflix with little choice but
to pay a “toll.” That might sound like a classic network neutrality
violation. But surprisingly, leading network neutrality proposals
wouldn’t affect this kind of agreement at all.
That’s because Comcast wasn’t technically offering Netflix a “fast
lane” on an existing connection. Instead, Netflix paid Comcast to accept
a
whole new connection. The terms of these agreements, known as
“peering,” have always been negotiated in an unregulated market, and
network neutrality regulations don’t apply to them.
In theory, Netflix’s deal with Comcast doesn’t violate network
neutrality because everyone on this new pipe (e.g. only Netflix) is
treated the same, just as everyone on the old, overloaded pipe gets
equal treatment. But it’s hard to see any practical difference between
the kind of “fast lane” agreement network neutrality supporters have
campaigned against and Netflix paying Comcast for a faster connection.
So why hasn’t interconnection been a bigger part of the network
neutrality debate? Until recently, it was unheard of for a residential
broadband provider like Comcast to demand payment to deliver traffic to
its own customers. Traditionally, residential broadband companies would
accept traffic from the largest global “backbone” networks such as Level
3 for free. So anyone could reach Comcast customers by routing their
traffic through a third network. That limited Comcast’s leverage.
But recently, the negotiating position of backbone providers has
weakened while the position of the largest residential ISPs — especially
Comcast, Verizon, and AT&T — has gotten stronger. As a consequence,
the network neutrality debate will be increasingly linked to the debate
over interconnection. Refusing to upgrade a slow link to a company is
functionally equivalent to configuring an Internet router to put the
company’s packets in a virtual slow lane. Regulations that try to
protect net neutrality without regulating the terms of interconnection
are going to be increasingly ineffective.
No comments:
Post a Comment