[60232] in North American Network Operators' Group
Inconsistant origins
daemon@ATHENA.MIT.EDU (Joe Provo)
Mon Aug 4 15:01:33 2003
Date: Mon, 4 Aug 2003 14:56:07 -0400
From: Joe Provo <nanog-post@rsuc.gweep.net>
To: nanog@merit.edu
Reply-To: nanog-post@rsuc.gweep.net
In-Reply-To: <Pine.LNX.4.44.0308042033080.26311-100000@uplift.swm.pp.se>
Errors-To: owner-nanog-outgoing@merit.edu
On Mon, Aug 04, 2003 at 08:35:14PM +0200, Mikael Abrahamsson wrote:
> On Mon, 4 Aug 2003, Mike Donahue wrote:
> > My company owns a class C, and we're switching ISPs. The "new"
> > provider is telling us that they can start announcing, without us
> > having to tell the old provider to stop announcing. I.e., a 2-3
> > day period where both are announcing our class C.
Inconsistant origins are bad. Which path is supposed to be believed?
Which path is supposed to be selected? From what point of view? None
of these are defined conditions; non-determinism is not good for
performance. Expect some things to work, some things to be strange,
and some to be lousy.
Your best bet is to minimize the overlap period.
Joe
--
RSUC / GweepNet / Spunk / FnB / Usenix / SAGE
