[107931] in North American Network Operators' Group
Re: Mechanisms for a multi-homed host to pick the best router
daemon@ATHENA.MIT.EDU (Valdis.Kletnieks@vt.edu)
Thu Sep 18 12:15:21 2008
To: Cayle Spandon <cayle.spandon@gmail.com>
In-Reply-To: Your message of "Wed, 17 Sep 2008 22:32:29 EDT."
<71c13900809171932m477b5d51jf0e0f9883546e3c3@mail.gmail.com>
From: Valdis.Kletnieks@vt.edu
Date: Thu, 18 Sep 2008 12:15:07 -0400
Cc: nanog@nanog.org
Errors-To: nanog-bounces@nanog.org
--==_Exmh_1221754507_3536P
Content-Type: text/plain; charset=us-ascii
On Wed, 17 Sep 2008 22:32:29 EDT, Cayle Spandon said:
> (Problem 2) If there is a topology change after the TCP connection has been
> established, the traffic might follow a sub-optimal path.
Another possibility is that the connection was originally established *during*
a link outage, so the initial part of the connection was done over a sub-optimal
path, and that the topology change puts it back to the normal better path...
A possible *biggger* issue is that "toss the reply packet back where the
original came from" makes traffic-engineering your outbound packets a lot more
challenging - you end up having to play announcement games upstream of your
N routers to engineer your *inbound* traffic so your outbound packets do what
you want.
--==_Exmh_1221754507_3536P
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Exmh version 2.5 07/13/2001
iD8DBQFI0n6LcC3lWbTT17ARAndHAJ0VIWrEt+ZSBuIuJNH+1Lpke7kY8gCfeVON
E2+V/Cvkg8N/BgQLqZEwg6g=
=ySRP
-----END PGP SIGNATURE-----
--==_Exmh_1221754507_3536P--
