BGP Multipath [Easy Cisco Configuration]

hello welcome to this right memo cutting UK tutorial on bgp multipath essentially BGP multipath is a way of low bouncing across two links with BGP many of the recent protocols do this by default but with BGP what you get is we have a room over here 10 dot 200 as a rosy row such 24 she's been advertised to Paris via both of these links be GPS default behavior is two dishes one of these links so Paris will just use either this link or the geek one link to to send its packets over to London the Destin's 10.2 hundred zero zero slash 24 network let's just show you that in operation first so I'll bring up the parish router and we'll take a look at the routing table so I've already got this company with PGP and we can see the only one link is being used yes as it gets the 10 dots 200 zero zero network I've grown by the 192 168 0 to link which is the bottom on there yeah so this link is being completely unutilized so you can see in the BGP summary it's up but it's not it's not actually being used for that particular send out to hundreds 0 0 especially for briefings ok so just to show you the failover so you can see 191 681 to become in the active neighbor I will paint the currently working link so just do that I shouldn't on the interface which is one on this route oh so he keeps a 0 on this route oh okay I see how that's gone down the neighbors come down here and if we look that meeting table now we can see that the other room is in use so the one mines he wants to say 1.2 so that's now the top link the sinews and the bottom link is is actually not being used at all she shut it down oh that's been that its face back up again and Papert I was the bond link is in your skin and something is just waiting there in case of failure let's get into the equation of multipath it's really simple you just go in the occurring ADP configuration is 100 in my case yep and we just do maximum paths and there we go everything should work fine from this point forward and both things should be in use take a look so we can only see one again that's trafficking in the PT 3 process so that brings down both neighbors brings it back up again sometimes we have to do this too to wait a few moments for us to repopulate the routing table and we can still see one neighbor inside inside here so I'll let you out a little bit the secret I deliberately put a route map on the top link say that those things are not equal cost the reason I did that is to show that an equal paths links will not be considered for load balancing so that's the reason why we've got just this particular neighbor here 191 6 8 0 2 as we know the bomb link that that is the only one that's being considered for BGP right now the top one can't be multipath because it's not equal cost so we can see what I'm talking about in a PDP trip so the local referenced I've set on the next hop 191 6 8 1.2 subset that's were local preface of 10 that it follows a hundred and higher is more preferred so we can see the little parrot symbol down here for the 10.2 hundred zero zero Network this is the preferred route so let's remove the local preference cough equation that I've on there and we should see that the load balancing begins to work because the parts are now equal cost right now okay clear the bgp process again it's not the kind of thing you do in a production environment but obviously this is just the gns3 lab so I'm quite happy to bring down the PDP neighbors there we go perfect we see two routes that 10.000 network and that is what it should look like when you're load balancing with eg P multiple we can take a look at a few different locations so that's what's doing the routing table and we're looking at PDP River and see what that looks like sleep 10200 network here we've got the first mix hop in the second next hop I see little hem there which is the nose that that's a multipath many people as well so that they've both been used for this particular next hop perhaps if a network sorry and we can also in the most important place really in the natural falling table itself and the fib I didn't ask IP safe and if we see in here it's the reaches there and you know we've got two different destinations there for that one prefix so that's a proof that we're balancing their rights I've done with the scenario example B I'm not going to pumping this one you know a lil not gonna conquer this one but just to kind of give you a bit more than understanding of how this might work in the real world as much as what valid configuration as the previous one was you might see this one more often where instead of just one London Rita we've got to London Rita's now maybe London's a large office and it requires requires redundancy essentially so if one reads fails another one that's back it up what we might see in this example is a network from the LAN we could use the same 110 200 0 0/24 would be advertised to London routers and then the London readers would advertise down both these separate links to the parish router and then if we have you know multipath enabled on the parent parish route oh it will work exactly the same as a previous example even though they're two different routes is it makes no odds to two BGP multipath it's just going to send one packet down King zero on one packet down keep one just the same as it would have done in the previous if we want a faulty path to work both ways then we'd have to configure it on one as well we only configured on Paris certainly routes that parish received would get the multipath treatment that's pretty much it I hope that helped you the fundamentals of equal cross load balancing using bgp multipath as with most things the explanation takes far longer than the configuration thanks for watching

Channel: writemem TV

Views: 2,804

Keywords: BGP, eBGP, routing, CCNP, Multipath, load balance, equal cost, cisco

Id: XWyCQVTMi8c

Length: 9min 52sec (592 seconds)

Published: Fri Aug 30 2019