From: Yinglam Cheung (ccie6961@yahoo.com)
Date: Thu Apr 29 2004 - 00:03:31 GMT-3
Yes, Cisco changed the default bgp behavior from synchronization to "no sync" one or two years ago. I noticed that after upgrading a GSR with an IOS version in ST train.
I believe Cisco has done this quietly on other IOS trains. :)
"Howard C. Berkowitz" <hcb@gettcomm.com> wrote:
At 9:39 AM -0700 4/28/04, Edwards, Andrew M wrote:
>I'm trying to understand methods to keep synchronization on in BGP and
>provide BGP to OSPF redistribution with route reflectors.
Perhaps someone can confirm this, but I seem to remember that Cisco
finally admitted synchronization is a completely obsolete technique
and has taken it out of the exams. The general idea of
synchronization, in part, comes from RFC 1403, which has been put in
"HISTORIC" statues -- i.e., obsolete -- by the IETF.
So, we have the problem, which may not be fully understood especially
by some review materials, that things like route reflectors were
never intended to work with synchronization, and developing practice
examples to do this is fundamentally futile. "synchronization has
been assimilated..." :-)
>
>You know the problem where the route reflector server receives an update
>from a route reflector client that is redistributing BGP to OSPF.
>
>When the route reflector server gets the update, it reflects that update
>to all other RR clients but changes the BGP ID to itself.
>
>Obviously the other BGP RR clients get the BGP update but the OSPF
>router ID and BGP ID do not match on the clients so the BGP route is not
>marked
>As a best path ">"
>
>So the question I have is what methods are available to make this work
>with synchronization on and using route reflectors?
See RFC3345 Border Gateway Protocol (BGP) Persistent Route
Oscillation Condition. D. McPherson, V. Gill, D. Walton, A. Retana.
August 2002 http://www.isi.edu/in-notes/rfc3345.txt
>
>Is the answer go to full mesh or transfer to confederations?
The above RFC generally recommends full mesh within a cluster, or
additional constraints on IGP metric and other things that affect
preference within the cluster. It's good to read this and compare
the ideas in it with the reasons OSPF always prefers intra-area
routes over inter-area over external.
>
>I'm stumped on how to change the BGP router-id to the originators BGP
>router ID on the RR server.
>
>Thanks for the input... or clearing up my confusion.
>
>Andy
>
>_______________________________________________________________________
>Please help support GroupStudy by purchasing your study materials from:
>http://shop.groupstudy.com
>
>Subscription information may be found at:
>http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Mon May 03 2004 - 19:48:57 GMT-3