International routing

mathis@pele.psc.edu Fri, 04 May 1990 21:06 UTC

Received: from devvax.tn.cornell.edu by NRI.NRI.Reston.VA.US id aa12438; 4 May 90 17:06 EDT
Received: by devvax.TN.CORNELL.EDU (5.59-1.12/1.3-Cornell-Theory-Center) id AA20212; Fri, 4 May 90 16:45:08 EDT
Received: from pele.psc.edu by devvax.TN.CORNELL.EDU (5.59-1.12/1.3-Cornell-Theory-Center) id AA20208; Fri, 4 May 90 16:45:03 EDT
Received: by pele.psc.edu (5.57/Ultrix2.4-C) (cf: do 2/28/90 --MM--) id AA27251; Fri, 4 May 90 16:44:11 EDT
Message-Id: <9005042044.AA27251@pele.psc.edu>
To: tewg@devvax.tn.cornell.edu
Cc: pscnet-admin@psc.edu
Subject: International routing
Date: Fri, 04 May 1990 16:44:09 -0400
From: mathis@pele.psc.edu

Consider the following single rule:

Forbid to listening to routes to any US network via any international link
landing in the US.

This should be a strong requirement: There are many examples of electronic
commodities bearing export restrictions.   What if non-exportable traffic
between two domestic locations uses an international path?

One affect of this is that international route churn CAN NOT affect
US domestic traffic.   Major win!!

Consider the following additional rules:  Forbid:
North American routes (US mainland, CAnet, Mexico) in either direction between
the pacific and Europe.
Pacific routes via links landing on the east coast of North America.
European routes via links landing on the west coast of North America.

These four rules will have the affect of limiting any route churn to a single
continent.  I claim that as a practical matter this will result in reasonable
routing because it forces symptoms of bad engineering to be relatively
local.   The people who will suffer the most will be the ones connected to 
broken international mid-levels/backbones.

We may also want to forbid announcing european routes back to europe without
explicit prior arangements.   This is fuzzy because in some cases the US may 
still be the best path.

--MM--