Sprint v. Cogent, some clarity & facts

Joe Greco jgreco at ns.sol.net
Mon Nov 3 10:28:48 CST 2008


> The FACT is that a point-source sending traffic to distributed  
> receivers combined with hot-potato routing puts more of the cost on  
> the receiver.  That fact is not in dispute, apparently even you agree.

Mmm, that's really not a fact.  I like the way you painted it though.

When you're looking at fundamental cost, let's not get confused by 
labelling it as "distributed" receivers.  There are several elements to 
the cost of getting data from point A to B.

Consider some examples.  Let's pretend:  You are a "point-source" sender 
on the east coast.  I am a "point-destination" receiver in the midwest.

The receiving network can build out its own national network, so that a
node is near you, and you (or someone that transits your packets to it)
peer with it.  In such a case, yes, hot-potato routing dumps most of the
cost on the receiving network, which is the sort of case that commonly 
exists these days.

However, there is no reason that receivers need to build national
networks.  A small receiver (think: regional ISP) could certainly bring
a line down to Chicago, get relatively inexpensive transit bandwidth,
and leave the sending site with the task of purchasing transit from 
someone who had a national network in order to get packets to the 
receiving network.

Now, there are a variety of problems here.  One problem is that there is
inherently a much higher cost-per-bit to actually deliver bits on a
residential broadband network than there is where bits are delivered in
a data center.  Despite this, residential broadband network feel a
downward pressure on price.  No successful residential broadband network 
actually has a network capable of delivering large numbers of bits to all
its customers simultaneously, and reliance on incredible overcommit is 
very common.

But there's another factor.  Many ISP's "feel" that it is cheaper to
internalize network costs; a regional network might still get a connection
to the east and west coasts, and peer on each coast, on the theory that
doing this was "cheaper" than paying transit.

In doing so, however, they internalize that cost for what-used-to-be-
transit, and remember that transit used to be paid for at both ends, by
the sender and the receiver.  Now it's all being paid for by the receiver,
at least in this example.

Therefore, I will take issue with the statement "puts more of the cost on
the receiver."  I will certainly agree that the receiver has underwritten
more of the cost, but as a deliberate choice.

This is, of course, only a very small component of a much more complex
picture, but it can be used to help people understand why things such as
paid peering exist.

In the case of Cogent, I have seen numerous examples of situations where
data entered the Cogent network and left it within the same building; one
has to assume that this is very inexpensive for Cogent, of course based on
the fact that their peering with other networks is settlement-free.  At 
least in major POP's such as Ashburn, this is actually more the norm than
not, last I was able to see live data from a customer to Cogent.  $10/meg 
from a paying customer, spit out onto other networks settlement-free, 
Cogent's costs are rather minimal there.

>  From that fact, you can argue whether that is grounds for de-peering,  
> settlements, etc.  But the fact stands.

No, it doesn't.  It's not even a fact.  It's just the way things often work
right now, because most have chosen to work it that way.

> Personally I think business problems have a business solution.  For  
> the ratio problems in 1998, Above.Net (and others, probably), agreed  
> to carry the traffic and deliver it closer to BBN's eyeballs, thereby  
> shifting the majority of the cost to AN.  Dave (Rand this time, not  
> Schaeffer) actually preferred it that way, saying he trusted his  
> network more than BBN's and AN's customers pay him for quality.

That would seem a reasonable solution.

Cold-potato routing is relatively difficult to do well on a large scale,
but perhaps a focus on this could reduce the "political" problem.  Cogent 
has been depressing the price of bandwidth for years, largely because of 
their hot-potato routing strategy.  Perhaps this isn't Cogent's fault so 
much as it is a failure to engineer equitable cost-management strategies 
that would discourage companies like Cogent from selling for $4/meg.

Please note that I'm mainly trying to paint a more accurate and
comprehensible picture for the audience here, Patrick.  I expect that
none of this is news to /you/.

... JG
-- 
Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net
"We call it the 'one bite at the apple' rule. Give me one chance [and] then I
won't contact you again." - Direct Marketing Ass'n position on e-mail spam(CNN)
With 24 million small businesses in the US alone, that's way too many apples.




More information about the NANOG mailing list