How to change routing protocol from RIP to OSPF without losing contivi

[brad]

  This is a question more about the process to do this task than specifics of configuring OSPF.

 I have an existing network that covers 7 remote locations using a private ASE cloud Ethernet service. Currently RIP is the routing protocol being used today. There is a project coming up to add redundant links to these locations so I cannot continue using RIP. OSPF is the new choice.

Hardware: Catalyst 9200 mostly 48 port POE switches running V17.3 with the advanced licensing, routing is on using RIP V2.

Config. The cloud ASE Ethernet service is the hub (Like a central switch) and the 7 remote locations are spokes. 

  The question is how do I implement OSPF without breaking the existing connectivity?  

I may be wrong, but I think if I turn off RIP at one of the remote location, I will lose connectivity to that remote location so I would have no remote way to turn on or configure OSPF.

I do not have the man power to have someone at each of the 7 location and each location is about 20 miles apart. Hitting all 7 location in a single day would be difficult.

Any ideas on how to implement this routing protocol change from the main location would be greatly appreciated.

[Harold Ritter · Accepted solution]

Hi @brad ,

Here's a simple and safe strategy that I have used in the past for routing protocol migration.

1. Add the ospf configuration to all sites, but with an admin distance of 121 (distance 121 under router ospf) to make it less preferred than rip

2. Make sure all the ospf neighbors are up from hub to spokes and that all LSA are being propagated properly using the

show ip ospf database

command.

3. Once you are ready to migrate, remove the

distance

command from all sites to make ospf preferred.

4. Check that all sites have all the routes being learnt via ospf.

5. Remove rip configuration from all sites.

Regards,

[brad]

Thank you Harold

That makes perfect sense. I did not see anything out there that definitively said I could run 2 routing protocols on the same device. I would not have thought about changing the “cost” of the link to control what route was used, RIP or OSPF. I was not ready to get an answer this quick. I am still in the process of replacing 16 c3650 that did not have the routing licenses with new c2900 with the advanced routing licenses. Once that is completed I will begin the OSPF migration phase so it will be a week or 2 before I can test.  I will update the post at that time!

Thanks again

Brad Moon

[Harold Ritter · Accepted solution]

You are welcome @brad . Just to make sure that we are on the same page, it is not the link cost that needs to be changed but rather the ospf administrative distance (AD). You can use 121 for the ospf AD, as the rip AD is 120.

router ospf xxx

distance 121

Regards,

[brad]

Thanks for the clarification. I was using “cost” in the most generic routing terms but other readers may not have not have noticed that I was doing that. My apologies to all that I may have confused.

Thanks

Brad Moon

[MHM Cisco World]

take me some time to check something, 
RIP vs OSPF with OSPF have AD large than RIP. 
that work but point need to be full understand here 
RIP v1 is classful compare with OSPF which is classless the router will add OSPF prefix in RIB even if it have higher AD than RIP ? why because it see two different prefix and that can lead to disconnect or loss connectivity. 
so you need 
first change the RIP to v2 
add no auto summary 
this make RIP advertise classless and hence compare to OSPF the RIP will win. 

NOTE:- friend try do lab before go with my solution. 
good luck 

[Harold Ritter]

Hi @MHM Cisco World ,

As per original message, he is already using rip v2.

Hardware: Catalyst 9200 mostly 48 port POE switches running V17.3 with the advanced licensing, routing is on using RIP V2.

Regards,

[brad]

  Sorry for the late update, but I had to get 16 new replacement switches installed in production before I could start the project. Original switches were C3650s

  With 5 of the new switches I built a test network to simulate what was in production. I then enables OSPF with AD set to 121 as Harold suggested. I was able to see all switches were communication with each other using the OSPF Neighbors and OSPF Adjacency. That let me see that all of the switches were talking to each other using OSPF. Show ip route however only displayed the RIP routes. My guess is that all of the routes were duplicated in both protocols and RIP was the most trusted route so that is all that was displayed. As soon as I set OSPF AD back to 110, the routes in the table immediately were displayed OSPF routes for that switch. The RIP routes were now gone. I was able to connect to each of the remote switches and remove the modifies AD one at a time until all switches were showing all routes as OSPF. No loss of connectivity, not even a drop from an extended ping that I had going at the time.  

  I now have all of the new switches installed and running RIP V2 as the original switches were doing. Next week I hope to make the switch to OSPF in the production network. I will keep you posted.

Many thanks to Harold! I was sweating this one.