[66808] in North American Network Operators' Group
Re: Outbound Route Optimization
daemon@ATHENA.MIT.EDU (Valdis.Kletnieks@vt.edu)
Mon Jan 26 10:31:43 2004
To: Michael.Dillon@radianz.com
Cc: nanog@merit.edu
In-Reply-To: Your message of "Mon, 26 Jan 2004 10:35:38 GMT."
<OFF3F89000.C0AF803A-ON80256E27.003897DD-80256E27.003A4D4D@radianz.com>
From: Valdis.Kletnieks@vt.edu
Date: Mon, 26 Jan 2004 10:30:38 -0500
Errors-To: owner-nanog-outgoing@merit.edu
--==_Exmh_-418178306P
Content-Type: text/plain; charset=us-ascii
On Mon, 26 Jan 2004 10:35:38 GMT, Michael.Dillon@radianz.com said:
> and such things may also be possible with MPLS. But are any of
> the researchers seriously looking at how to provide a network
> in which all packets flow through two diverse paths to provide
> better reliability?
There's enough banana-eaters that can't get their heads wrapped around
the way BGP does things. How many "I have a routing issue" postings
are we going to see with a protocol that splits up packet streams?
I'm going to assert that *in the aggregate*, network optimality is higher when
there's only one path, because when it dies it gets fixed. If you have two
paths, and one is lossy/congested/etc, the resulting jitter and retransmits
will (a) make things worse and (b) be too much for the banana eaters.
Yes, we can probably make something "better" than BGP. But will we
be able to understand it?
--==_Exmh_-418178306P
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)
Comment: Exmh version 2.5 07/13/2001
iD8DBQFAFTKecC3lWbTT17ARAr6DAJ4zDq+3oEAbN5U5YlKATAWkXbkY1QCg953Q
lJdsPxR3nPPkTPlgGkqvyD0=
=7e+L
-----END PGP SIGNATURE-----
--==_Exmh_-418178306P--
