BGP propagation Q

[pnschurr]

Hi,

Question re iBGP and eBGP. Let's say I take some eBGP routes into my AS.

I then pass them across the AS via iBGP.

On the other side, I have them as iBGP and propagate them onwards downstream via eBGP (assuming all the normal stuff is ok, as_path etc).

On that outgoing router, does the prefix need to be in the actual routing table as iBGP for it to be propagated by eBGP? or will BGP see it as the preferred path in the BGP table, and then propagate it - regardless of what is in the route table for that prefix.

I believe BGP propagates prefixes based on the BGP table, not the route table.

Is that correct?

thanks

peter

ps. not talking any redist to/from any other protocols.

[pkhatri]

Hi,

A BGP router will only advertise BGP routes that it itself uses. Therefore, the iBGP routes need to meet 2 conditions in order to be further propagated:

- it needs to be the 'best' route in the BGP table

- it needs to be installed in the routing table. There are cases when the best BGP route is not in the routing table due to the presence of the route from another, more preferred source.

Hope that helps - pls do rate the post if it does.

Paresh

- th

[Georg Pauwen]

Hello,

in addition to Paresh's post, there are actually ways to originate a prefix into a BGP table without the corresponding match in the routing table. BGP Conditional Route Injection is such a feature (see the link below). The idea is to have BGP originate specific routes based on the existence of another route. This is mainly used in conjunction with route aggregation, which obscures more specific routes.

BGP Conditional Route Injection

http://www.cisco.com/en/US/products/sw/iosswrel/ps1839/products_feature_guide09186a008056c862.html

Regards,

G