Determine Traffic Path in BGP

[CiscoNutt]

Lets say we have a routerA that has 2 paths to routerE.  One path goes routerA --> routerB --> routerC --> routerE. The other path goes routerA --> routerD --> routerE.  Assume all metrics are default value.  BGP is configured with an MPLS VRF running over it.  In addition there is a GRE tunnel that connects routerA to routerE.  In show ip route vrf VRF-NAME the next hop is shown as the GRE tunnel interface on routerE.

I am assuming that the path the tunneled traffic will take is the shortest path through routerD, because of the shortest AS path...or is my assumption wrong?

[Raju Sekharan]

Hi

the number of AS involved is not very clear from your explanation. But if you have not modified local-preference and weight, then path with shortest AS lengh is the preferred path

Thanks

Raju

[CiscoNutt]

There is only one AS involved.

[Raju Sekharan]

Hi

if all are in same AS, then the length of AS is going to be same for same prefix on all routers in that AS

Thanks

Raju

[CiscoNutt]

Then what is the deciding factor (given that all values are set to default) on which of the two paths are chosen? or will the traffic be loadbalanced?

Thank you for your replies.

[Raju Sekharan]

it will be based on BGP best path selection Algorithm

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094431.shtml

Thanks

Raju

[blau grana]

Assuming everything has default value, link bandwith between routers in your AS is same, BGP will decide based on 8th step, which is IGP metric.

If you have access to routers, enter sh ip route IP or sh ip cef IP also traceroute IP to confirm. IP is IP address of next-hop (probably PE router) from output of sh ip route vrf VRF.

[CiscoNutt]

sorry forgot to mention a few things.

IGP running is ISIS with fiber connecting the routers directly together.

Traceroute shows the next hop as being routerE as there is a tunnel connecting the two routers.  Same for show ip route vrf VRF-NAME shows the next hop as the tunnel interface on routerE.

so the path is chosen based on the ISIS metrics.  Yet the ISIS shows the next hop as being routerE, I am assuming this is because of the tunnel.

[blau grana]

I assume that GRE tunnel endpoints are in global routing table and ISIS is running accross GRE tunnel. Try traceroute from both ends of GRE tunnel to confirm path which GRE packets take.

# traceroute GRE_TUNNEL_DESTINATION update-source GRE_TUNNEL_SOURCE

As I wrote earlier, if BGP parameters are at their defaults, IGP cost will decide which path will be choosen, in this case ISIS.