cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
3611
Views
5
Helpful
19
Replies

Best Path BGP for dual link between two AS - how force only one path

_|brt.drml|_
Level 1
Level 1

Question:

In BGP I understand how to influence routing information comming from my peers. However, every example I have learned in the course was , dual-homed... 

However, now I stumble on this situation.

I have two routers, both connected with two 'links'. Both router are different AS.

In my BGP table I receive routers from my AS.

 

I emulated one link as a low bandwidth and high delay link. 

I emulated one link as my high speed link.

 

What is the best option in BGP to influence the route selection? 

 

I thought local-preference or med, but this only influence the route I install and use... It does not influence my 'interface' decision 

Any ideas are welcome. 

 

Thank you.

 

B

19 Replies 19

_|brt.drml|_
Level 1
Level 1

So what worked in the end.

We created a routing map that captures the traffic with an ACL for that particular network.

This route-map sets 'interface' to the tunnel we like.

 

If the tunnel is known in the ip routing table it shall send traffic via the known interface. 

you can check this with 'show ip cef vrf *** 'ip''. why!: when the tunnel is down, then it does not appear in the global routing table. This gave an unwanted side affect for this traffic. All other traffic is diverted except the captured traffic that is 'set to interface'. via the route map. 

 

All other metrics like suggested like MED, did not change traffic via a particular Interface...  

 

Personally I do not like the solution. With an applet I can kick out the route-map on the vlan interface.

 

in short 

 

access-list standard 1
 permit 'your vlan or other IP traffic' any
 deny all

route-map SELECT_TRAFFIC_SET_INTERFACE permit 10
  match ip address 1
  set SLOW_INTERFACE

in our situation: vrf vlan with MP-BGP same AS
interface vlan****
 ip policy route-map SELECT_TRAFFIC_SET_INTERFACE

verification:
clear counter
ping repeat 'many' 
show policy-map interface ***** -> 'if any'
show ip interface *****

 

I hope that this was interesting. 

Thank you for the guidelines. I especially liked the DMZ_BW in BGP. Never heard from it. Wasn't even in my Cisco BGP training. 

 

The overall issue; training provides multipath, but always between different AS... 

 

 

Hello
So you went with policy based routing and not to utilize bgp attributes such as weight and as-prepend.
Don't forget as I stated previously with PBR this will ONLY effect egress traffic you will still need to make sure return traffic is ALSO coming in via your primary link.


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

Indeed, it does ask for many attention. We come from one problem to another. Such as our eigrp metric. The slowest link is one hop the fastest many... but that is solvable.

The other problem is much more difficult.



My ICMP does leave the router and follows the PBR intention.

However, now we lose contact with wireless controllers. Beats me how and why, guess that this is another problem that suddenly appears...

Even when we disable the PBR, the problem stays. Oh, it's so fun








I don't believe that the solution for any routing issue is using PBR..
as You mention "Personally I do not like the solution." me too

using PBR is ok for special case but always using it...NO 

Yep, but if we like to disable this, then I and my colleague need to redesign our main central routers.

I guess all the metrics are in GP are perfect for between different AS. In our case we have the same AS, the connecting interfaces are dynamically.

So using next-hop is not so easy. But now I think of it.

If an applet could read the connecting router for my slowest link, then I can configure next hop for a specific route. This could help





B