Excerpts from Melinda Shore at 17:55:57 -0400 on Wed 21 May 2008:
On 5/21/08 5:49 PM, "eburger(_at_)standardstrack(_dot_)com"
So we have reinvented STUN?
This is probably closer to Paul Francis's NUTSS stuff without
the cool crankback and especially without resolving the location
With NUTSS the destination is (probably) seen in routing from the
start, but signaling is required to set up a desirable path. Here the
problem is that neither source nor destination is globally routed.
I'm concerned about scaling. That's a lot of dynamic state in the
core routers, precisely where we don't want it. He's probably better
off source-routing. See NIMROD
<http://ana-3.lcs.mit.edu/~jnc/nimrod/docs.html> and remember what it
stands for! :-)
He should also see the proposals in the Routing Research Group
IETF mailing list