To influence traffic you originate towards your upstream you can do it by preferring received prefixes with weight (local to the router), local preference (local to your asn), as path, med among other things. Obviously communities to set any of the above will work too assuming you guys pass communities between eachother.
If you want to influence traffic towards your asn the options are prefix length, as path, meds and influencing your upstream's internal policies using communities (if allowed, etc). Announcing less than RIR allocation towards your upstream can be effective sometimes. If you have a 10/16 and announce it as 10/16 on both links _and_ 10/17 on rtr A, 10.0.128/17 on rtr B you've effectively just split your traffic (assuming even distribution of traffic to every IP address ;-)). Tag the more specifics as no-export - the guys on the other side of the planet don't need to know about your /17s as the /16 will cover them and only your upstream - to which you're dual homed to - needs to be aware of them. Assumes that your upstream will let you do this... Most will up to a certain prefix length, like /24. Local preference is by default 100. On the router you want to prefer routes on create a route-map which somehow matches the prefixes you wish to pref/depref & attach that route-map to the session in inbound direction. You can match on the prefix itself, a community or communities added to it or the as path. Raise the local preference to 110 (or whatever as long as it's > 100) and you're set for those prefixes. You can lower the local preference this way too. ip as-path access-list 4 permit ^42_42_666_ ip community-list standard quux permit 42:4242 /* LP=110 on routes tagged 42:4242 */ route-map map-foo permit 10 match community quux set local-preference 110 /* LP=90 on paths starting 42 42 666 and anything after */ route-map map-foo permit 20 match as-path 4 set local-preference 90 If you're doing several match clauses within a route-map section all the match criteria have to be met in order to have the set statements applied. You need to at least soft clear inbound (or trigger route refresh) to get the policy applied. Whether this actually does anything else depends on how your asn is setup - if there's ibgp everywhere the above will be propagated. If you're running bgp only on A & B and originating a default with different metrics on each, traffic with no specific routes will always go towards the best router originating 0/0 which then will send traffic to the other (assuming A is has best 0/0 and B has best path to some prefix). Kaj > From: Taqdir Singh <[email protected]> > Date: Sat, 15 Aug 2009 01:10:28 -0700 > To: <[email protected]> > Subject: [OSL | CCIE_RS] BGP Local preference for local AS > > > > > I have two exit point routers ( A and B) , one to ATT and other to Verizon. > Both routers are in AS 100. > > I want to use loadbalancing. > > Can I do something on routers A and B using BGP so that for the some subnets > inside my local AS 100,, I can make next hop of A and for remaining subnets I > can make next hop of B. > Please note I only want to configure on router A and B. other question is... > Can we use local preference for local AS ? > > Sincerely, > Taqdir Singh > 91-9911709496 > > > Do today what others won't, So that you can live tomorrow what others can't > > > > > > See the Web's breaking stories, chosen by people like you. Check out Yahoo! > Buzz <http://in.rd.yahoo.com/tagline_buzz_1/*http://in.buzz.yahoo.com/> .
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ For more information regarding industry leading CCIE Lab training, please visit www.ipexpert.com
