UUNET Pulling Peering Agreements & replacing them with charging under non-disclosure?

Michael Shields shields at crosslink.net
Sun May 4 05:01:16 UTC 1997


> Or, it could be looked at as equal amounts of traffic. Right now uunet
> pulls 5 times as much data from my network as my customers pull from their
> network.

This same argument comes up every time!  We all wish there could be
nice clean technical criteria but there isn't any way to measure
*value* except by asking the customers, which in practice seems to be
done by unplugging the links and seeing who gets more complaints.
Attempts to require X network size are just a crude way to codify the
size "large enough that *my* customers will be unhappy if I'm not able
to reach you".

The value of connectivity from UUNET to any given network -- i.e., the
desire of UUNET customers to reach ISP-X customers minus the desire of
ISP-X customers to reach UUNET customers -- varies widely in magnitude
and even sign, so I think it's reasonable of them to negotiate the
number on a case-by-case basis instead of only free peering or else
paid transit from the price schedule.  Everyone seems to be upset
because the outcome is uncertain, less clear-cut than the status quo.
But if you think about it, there are many cases in which the value
*almost* justifies free peering, and if a bit of cash exchanges hands
then the table tilts -- in favor of better connectivity for everyone.

And I'm sure the NDA is just to avoid ISP-X whining "why should we pay
$N when ISP-Y didn't pay nearly that much" and the subsequent
explanation that ISP-Y adds value by hosting Very Popular Site.  They
could explain this n^2 times, or they could just require NDAs.  There
will be less gossip-fighting in the latter case.

It cannot be a case of an evil giant trying to dominate the industry
because UUNET isn't so big that they can sway the entire business
model singlehandledly.  UUNET must have set this policy because they
feel it's in their interest.  And I can see their point.  It *does*
make sense.
-- 
Shields, CrossLink.
[neither customer nor peer]






More information about the NANOG mailing list