From: Budak Timuralp (budak.timuralp@sabancitelekom.com)
Date: Wed Jul 23 2003 - 09:32:48 GMT-3
The question below was asked before by Carlos , but didnt get an answer.
While redistribution takes place in both directions from bgp to isis and
also from isis to bgp, the question below makes sense. In real life you
dont implement such configurations but it drives me crazy. I have build
this case, it works but i wanna know how?
Any help will be appreciated.
Wed, 28 Nov 2001
This question relates to the case set out in Doyle vol page 219. The
typology is set out on page 212.
Apologies to those that don't have Doyle on hand.
The facts are as follows
1) As 30 is multihomed
2) The 2 edge devices in as 30 are not peered i.e. no iBgp in as 30
3) The 2 edge routers redis their BGP routes into IGP - ISIS in this
case
4) The 2 edge routers redis IGP into BGP
The net result is that at least one of the edge devices in AS 30
will/could
feedback routes learned from other AS back into BGP showing AS 30 as
the
originating router.
On page 220 in example 3-84 Doyle shows that this feedback can be
avoided
by applying a filter list ensuring that AS30 only advertises routes
originating from the AS to other peers i.e." ip as-path access-list 1
permit
^$ "
My question is how will this avoid the feedback problem , all the path
info
is lost when the BGP redistributes into ISIS , when the "external
routes
are redis back into BGP at one of the edge devices in AS30 they show as
having originated internally and therefore slip past the filter list.
Tx
Carlos
Best regards,
Budak Timuralp
This archive was generated by hypermail 2.1.4 : Wed Aug 06 2003 - 06:52:50 GMT-3