Re: BGP4 questions
John Krawczyk <jkrawczy@baynetworks.com> Fri, 26 July 1996 13:27 UTC
Received: from ietf.org by ietf.org id aa03411; 26 Jul 96 9:27 EDT
Received: from cnri by ietf.org id aa03407; 26 Jul 96 9:27 EDT
Received: from p-o.ans.net by CNRI.Reston.VA.US id aa10969; 26 Jul 96 9:27 EDT
Received: (from majordom@localhost) by p-o.ans.net (8.7.5/8.7.3) id NAA16671
for bgp-outgoing; Fri, 26 Jul 1996 13:21:14 GMT
X-Authentication-Warning: p-o.ans.net: majordom set sender to bgp-owner using
-f
Date: Fri, 26 Jul 96 09:21:05 EDT
Message-Id: <9607261321.AA25137@pobox.BayNetworks.com>
Sender: ietf-archive-request@ietf.org
From: John Krawczyk <jkrawczy@baynetworks.com>
To: bgp@ans.net
Subject: Re: BGP4 questions
In-Reply-To: <199607251855.OAA03022@rasarit.midnight.com>
References: <199607251855.OAA03022@rasarit.midnight.com>
X-Orig-Sender: bgp-owner@ans.net
Precedence: bulk
Reply-To: bgp@ans.net
On Thu, 25 July, Cristina Radulescu-Banu (cristina@midnight.com) wrote: Hi. I have 2 questions: 1. What does this means: A BGP speaker needs to be able to support disabling advertisement of external border routers. (RFC 1771, page 22) 2. Did someone actually implement this? Cristina, I had to look at the spec to put this into context. This is in the section talking about the NEXT_HOP path attribute, and concerns what is sometimes referred to as "third party" advertisement. That is, if routers A, B, and C are all on the same multi-access network, A can advertise B's NLRI to C using "B" as the NEXT_HOP, so as to avoid sending packets out onto the same subnet twice (C->B instead of C->A->B). This is the default behavior in our implementation. I can think of two interpretations for "disabling" this: 1) Have A use its own address as NEXT_HOP. This would be necessary over an NBMA network where there is no direct connectivity between B and C. I can't think of any other good reason for doing it (that doesn't mean that there aren't other good reasons... :). 2) Prevent A from advertising B's routes to C. You might want to do this to save memory, but you are also sacrificing a backup mechanism for forwarding from C to B. You can do all of this in our implementation through BGP policies. It's not as easy as just turning a knob to say "don't do that". -jj
- Re: BGP4 questions John Krawczyk