[c-nsp] IPv6 BGP on IOS-XR

[Nick Hilliard]

On 28/07/2011 19:33, Lee Starnes wrote:
> Does anyone have any recommended reading for doing IPv6 BGP peering on
> IOS-XR? We setup our peer with Cogent and can get full routes in, but can't
> seem to figure out how to announce our /32. I had assumed it setup just the
> same way as IPv4, but can't seem to get it to announce out. I did some
> searching on Cisco's site, but didn't really find anything that pertained to
> how/what to announce and what not to announce.

bgp in xr is different from vanilla ios in that by default, it applies an
implicit deny on outbound prefixes.  Regular IOS applies an implicit
permit, which leads to prefix leakages.  XR definitely takes a more
sensible approach here.

What you need to do is to set up a route-policy for your neighbour, and
hook that route policy to whatever mechanism you want for prefix
redistribution.  Let's say in this example you want to hook it into a
static prefix list:

prefix-set my-ipv6-prefixes
 2001:db8:1234::/48
end-set

route-policy my-ipv6-policy
 if destination in my-ipv6-prefixes then
  pass
 else
  drop
 endif
end-policy

router bgp 65432
 neighbor 2001:db8:ffff::1
  address-family ipv6 unicast
   route-policy my-ipv6-policy out
[...]

Obviously, you can hook the route-policy into any filtering mechanism you want.

I am a total fan of XR's RPL.  It makes regular IOS route-maps look
lame-ass in comparison.

Nick