[c-nsp] Persistent Convergence Oscillation in BGP Confederations

[Brian Feeny]

I was wondering if anyone with experience in dealing with PCO in a  
Confederation, can share
what they have done to mitigate it.  I realize there are many ways to  
handle this, and I have read
RFC 3345 as well as alot of stuff out there, but knowing what others  
have used in a actual
operational environment would be interesting.

I plan to use a hub-spoke topology from a logical point of view for  
connecting sub-AS's.  Also
all the transit goes into the backbone/hub sub-AS (at least for now).   
I believe that alone would
stop type II churn if I read correctly.  It would seem you almost  
always need to raise the inter-sub-AS
IGP metric above the intra-sub-AS IGP metric to ensure loop free  
topology.

Brian

------------------------------------------------------------------------ 
------
Brian Feeny, CCIE #8036, CISSP    	e: signal at shreve.net
Network Engineer           			p: 318.213.4709
ShreveNet Inc.             			f: 318.221.6612