AS Path Prepending on XE and XR

This short article explores what happens when you do AS path prepending and apply it outbound to an eBGP neighbor.

Consider the following topology:

R1 originates 10.10.10.10/32 into BGP. R2 uses the following route-map and applies it to the neighbor statement for R3 outbound:

route-map AS_PREPEND
 set as-path prepend 100
!
router bgp 65001
 neighbor <R3> route-map AS_PREPEND out

What is the AS path for 10.10.10.10/32 as seen on R3? There are two options that might make sense:

1. 100 65001 65000 i

2. 65001 100 65000 i

The answer is 65001 100 65000 i. This is because the route-map is applied before the local AS is finally added to the AS path, when advertising the route to the eBGP peer.

R3#show bgp ipv4 uni | be Network
     Network          Next Hop            Metric LocPrf Weight Path
 *>   10.10.10.10/32   10.10.12.1                             0 65001 100 65000 i

Note that this is logic is not applicable when you change the next-hop IP address in the route-map. The router doesn’t then change this to its interface address used for the eBGP session. The next-hop that the route-map sets takes precedence.

AS Path Prepending on IOS-XR

This same logic applies when preforming AS path prepending on XR with the following route-policy:

#R2
route-policy AS_PREPEND
 prepend as-path 100
end-policy
!
router bgp 65001
 neighbor <R3>
  address-family ipv4 unicast
   route-policy AS_PREPEND out

The local AS is prepended after the RPL prepending takes place:

RP/0/0/CPU0:R3#show bgp ipv4 uni | be Network
Thu Dec  8 04:07:53.213 UTC
   Network            Next Hop            Metric LocPrf Weight Path
*> 10.10.10.10/32     10.2.3.2                               0 65001 100 65000 i