Asian assignment

William Allen Simpson <bill.simpson@um.cc.umich.edu> Mon, 24 May 1993 00:10 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa12231; 23 May 93 20:10 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa12227; 23 May 93 20:10 EDT
Received: from thumper.bellcore.com by CNRI.Reston.VA.US id aa11606; 23 May 93 20:10 EDT
Received: by thumper.bellcore.com (4.1/4.7) id <AA22520> for ietf-archive@nri.reston.va.us; Sun, 23 May 93 20:06:12 EDT
Received: from vela.acs.oakland.edu by thumper.bellcore.com (4.1/4.7) id <AA22512> for /usr/lib/sendmail -oi -fowner-pip X-pip; Sun, 23 May 93 20:06:11 EDT
Received: from via.ws07.merit.edu by vela.acs.oakland.edu with SMTP id AA23858 (5.65c+/IDA-1.4.4); Sun, 23 May 1993 20:01:13 -0400
Date: Sun, 23 May 93 12:57:35 EDT
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: William Allen Simpson <bill.simpson@um.cc.umich.edu>
Message-Id: <1205.bill.simpson@um.cc.umich.edu>
To: Dennis Ferguson <dennis@ans.net>
Cc: pip@thumper.bellcore.com, sip@caldera.usc.edu, tuba@lanl.gov
Reply-To: bsimpson@morningstar.com
Subject: Asian assignment

> Somehow your response leaves me feeling a bit uncomfortable.  I would
> point out that the current path from places in Thailand to places in
> Singapore appears to pass through Virginia and New Jersey, e.g.
>
> I can hardly wait to see where the packets go when people in Myanmar,
> or Viet Nam, decide they'd be best served by dealing with a European
> provider.
>
Thank you for a concrete example, and thank you for making
(inadvertantly, I'm sure) the case for metropolitan assignment.

You cited 2 Asian countries have (indirect) internet links to 2 U.S.
providers.  Because of the silliness of AN_S (yet another provider)
policy routing, all of their traffic goes through AN_S routers on the
East Coast.

Could you imagine the problems in those sites if they were assigned
based on provider, and then changed to a local provider as they grew?
Entire countries to be renumbered.

Worse yet, assume those sites have the usual assignment strategy
(first come, first served).  As their internet grows, they end up with
the hodge-podge mess we currently have in the U.S.

Using early metro-based assignment, future assignments will aggregate
naturally as connectivity increases.  Provider-based assignment requires
us to decide, years in advance, what the overall topology will be.

Or is your point that we should all have AN_S provider addresses,
because most of the world currently routes through you?  Thus ensuring
that it always will?

Bill.Simpson@um.cc.umich.edu