[c-nsp] partial prefix load in ibgp

[Gert Doering]

Hi,

On Wed, Oct 26, 2011 at 12:06:31PM +0400, Nikolay Shopik wrote:
> I've got two borders connected via ibgp, both receving full-view via 
> uplinks, and I notice one of borders receiving about 190K prefixes via 
> ibgp. 

This is a FAQ... BGP only forward-announces the best prefix it knows.

So if you have:

ExtA - Ra - Rb - Extb

and "Ra" thinks that 150k prefixes are "best via Rb/ExtB", then it will only
announce the rest (190k via ExtA) to Rb.

Prefixes preferred over Rb will never be announced back to Rb, and since
BGP will only ever announce the *best* path, you'll see less-than-full
tables on Rb.

(There is research going on to extend BGP here, but that's "the default
behaviour" and I'm not sure whether there's already implementations 
out there that support announcing multiple paths)

gert
-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert at greenie.muc.de
fax: +49-89-35655025                        gert at net.informatik.tu-muenchen.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 305 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20111026/dd98cd1d/attachment.pgp>