7

If you have a router which does not support OSPF, but your routing topology makes use of OSPF as all other devices do support it, what is the best way to get this routers networks in to the OSPF area?

My initial thought was to put a static route on every other router, but if there are multiple networks behind this router this could become very cumbersome very quickly.

This problem arises from the answer on another question: How do you get an ASA to announce NAT'd 'outside' addresses in to an OSPF zone?

4 Answers 4

12

What does the topology look like? How are you doing routing today towards these networks? Or is the router being deployed now? Do you have an OSPF capable router connected to this router that is not running OSPF?

If you do then I suggest you redistribute static on this router. Something like:

ip prefix-list static-routes permit 10.0.0.0/24
ip prefix-list static-routes permit 10.0.10.0/24
ip prefix-list static-routes permit 10.0.20.0/24
route-map static-allowed permit 10
match ip address prefix-list static-routes
route-map static-allowed deny 20

The explicit deny at the end is not really needed as there is an implicit deny but it makes the logic a bit more clear.

Then under the OSPF process:

router ospf x
redistribute static subnets route-map static-allowed
5
  • Would it be acceptable to run a cheap router that supports OSPF purely to redistribute static routes in to an OSPF area? I guess this would act a bit like a BGP route server. May 13, 2013 at 21:12
  • edited to add use case May 13, 2013 at 21:14
  • @Simon that would work but what a kludge. Why not replace the router entirely then. Everything not consumer grade does OSPF.... even going as low as Cisco 800s, Mikrotiks etc. May 16, 2013 at 4:23
  • @wintermute000 see the explanation in the question linked at the bottom of this question. The device can do OSPF, it's to do with NAT behind that device as its not a connected network. May 16, 2013 at 6:21
  • Then as long as you only have one point of connection to the NAT device (via the OSPF capable device), statics with redistribute static subnets. If two points of connection, google tag and filter, and use metrics or E1 vs E2 to set priority May 20, 2013 at 4:18
9

At the point where the non-OSPF domain touches the OSPF domain I would setup static routes and then redistribute those static routes into OSPF.

This ofcourse is a very static (no pun intended) setup, there is always the other option of running another (supported) protocol down to the non-OSPF device and then redistribute between the two protocols.

Have you got a use case where you would want to be doing this?

1
  • edited to add the use case May 13, 2013 at 21:34
3

The best way to do this would be through redistribution. If the non-OSPF devices do support RIP or EIGRP you could create a neighborship with the OSPF device and non-OSPF device and redistribute the routes into OSPF - and the rest of your network. This is relatively safe if the non-OSPF device is single-homed. If dual-homed then you'll need to beware of the possibility of routing loops due to lost metric when redistributing.

If running a different IGP on the non-OSPF device is not an option then I'm afraid you'll have to use static routes - be it manually put in or automated with some sort of script.

1
0

As other people have already pointed out, the only semi-sane design is to configure static routes pointing to (or across) the non-OSPF device at the edge OSPF device and redistribute them into OSPF. Configuring static routes on other OSPF devices would be a nightmare.

Now, assuming the non-OSPF device is totally IGP-clueless (it cannot run RIP or EIGRP, for example), there are two ways to make the routing at least a bit more dynamic (to detect failures in the IGP-clueless device, for example):

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.