Yeah, that's essentially what cisco said, though I'd guess that you ought to be able to at least not propagate the withdraw back to the peer who sent it to you... But that's just my opinion. Steve R./Merit ==

From nanog-owner@merit.edu Fri Jun 21 11:26:41 1996 Received: from merit.edu (merit.edu [35.1.1.42]) by home.merit.edu (8.7.5/merit-2.0) with ESMTP id LAA03178; Fri, 21 Jun 1996 11:26:39 -0400 (EDT) Received: (from daemon@localhost) by merit.edu (8.7.5/merit-2.0) id LAA16611 for nanog-outgoing; Fri, 21 Jun 1996 11:20:52 -0400 (EDT) Received: from chops.icp.net (chops.icp.net [199.0.55.71]) by merit.edu (8.7.5/merit-2.0) with ESMTP id LAA16603 for <nanog@merit.edu>; Fri, 21 Jun 1996 11:20:31 -0400 (EDT) Received: by chops.icp.net id <20689>; Fri, 21 Jun 1996 11:20:23 +0100 From: Sean Doran <smd@icp.net> To: justin@erols.com, nanog@merit.edu Subject: Re: Withdrawls and announcements attempt 2 Message-Id: <96Jun21.112023+0100_edt.20689+60@chops.icp.net> Date: Fri, 21 Jun 1996 11:20:18 +0100 Sender: owner-nanog@merit.edu Precedence: bulk Status: R

Keeping track of the state of who got announced what is likely to be a very very very bad idea for busy BGP talkers carrying today's amount of NLRI and instability.

There are some hacks around the simplistic "if it's in my RIB, I have to propagate withdrawals to all my neighbours" for some cases, but a more comprehensive fix would require some Thinking.

This should probably get migrated over to the BGP list.

Sean.