From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Mon May 24 2004 - 02:30:40 GMT-3
I will preface these marks by saying that I am answering from a
standpoint of best current practices in the ISP world, which doesn't
necessarily reflect what Cisco wants on the lab.
At 9:52 PM -0700 5/23/04, Ahmed Mustafa wrote:
>1) Is it true that redistributing bgp into ospf should be avoided as much as
>possible.
In general, other than very, very tightly controlled redistribution
of summarized or selected routes, it is a bad idea to redistribute
more than a few BGP routes into an IGP. IGPs were designed for fast
convergence with a relatively small number of routes, where BGP is
optimized for handling very large number of routes. There is a
never-implemented part of the OSPF specification, the database
overflow LSA, which was intended to help solve the problems we now
solve with iBGP.
In like manner, in the real world, it is not conducive to
redistribute IGPs into public BGPs. The normal practice, in the
interest of stability, is to redistribute blackboles or conceivable
aggregates.
>
>2) If the requirement leads towards using Confederation, but the ASs are not
>given. Can the AS from the private ASs be taken. In other words private AS
>will work with confederations.
>
Confederations made of one registered AS and all the rest private AS
is the most common case.
This archive was generated by hypermail 2.1.4 : Wed Jun 02 2004 - 11:12:16 GMT-3