Change cost for ABR summary route to non-backbone area on IOS-XR

[Lubos Lontos]

Hi,

I have network with Cisco 760X(ABR) and ASR9k(ABR) and I try to change cost for summary route on ABR to non-backbone area.

Configuration on IOS:

router ospf 100
 router-id 1.1.1.3
 auto-cost reference-bandwidth 100000
 area 0 range 1.1.1.1 255.255.255.255 cost 1000
 area 0 range 1.1.1.4 255.255.255.255 cost 1000
 area 0 range 10.0.0.0 255.255.248.0
 area 11 range 10.0.11.0 255.255.255.0
 area 11 filter-list prefix PL-Ar11 in

On IOS-XR(5.3.4) I can only filter prefixes with route-policy, but can't change cost for summary route. When I try to change ospf-metric in route-policy:

route-policy RP-Ar11-TEST
 if destination in PS-LooR3 then
  set ospf-metric 1000
 endif
 if destination in PS-Ar11 then
  pass
 else
  drop
 endif
end-policy
!

it can't be applied to area, because ospf-metric is not allowed in area route-policy.

On area range command there is no cost(metric) option:

RP/0/RSP0/CPU0:R1(config-ospf-ar)#range 1.1.1.3/32 ?
 advertise            Advertise this range (default)
 not-advertise     DoNotAdvertise this range
 <cr>
RP/0/RSP0/CPU0:R1(config-ospf-ar)#range 1.1.1.3/32 advertise ?
 <cr>

Any idea?

Thanks

Lubos

[xthuijs]

yeah that is not there in IOS-XR... one option would be to create a static null route for the summary and do a redistribute with a route-policy to adjust the cost. is that acceptable?

although it makes the route become an E1/E2 obviously.

cheers!

xander

[Lubos Lontos]

Unfortunately no,

I need change cost in exact route /32, if I make /32 route to null on ABR, ABR drop all traffic to this route because static route AD is 1 and OSPF 110. If I change AD for static route (more than 110) it will be not in rib and there will be no redistribution. I tried it before.

I need exact route because of label binding.

Thanks

L.Lontos

[Adam Vitkovsky]

I see,

so you’ve been doing a bit of traffic-engineering at the area boundary, and it turns out the feature is not available in XR,

Any particular reason behind areas in you MPLS backbone design?

How I usually do it (and I hate the idea of areas in MPLs core so I always run separate IGP domains instead) is, I use hierarchy using RFC3107 and hierarchical design.

You can either use BGP only at the area boundary and redistribute back to OSPF or use BGP end to end, I prefer the latter cause that way I don’t need to carry routes from other areas in OSPF/ISIS.

BGP at the area boundary gives you all the control you need over what prefixes are exchanged and with what metric.

adam

netconsultings.com

::carrier-class solutions for the telecommunications industry::