Just a very quick post tonight regarding Redistribution of OSPF into RIP and back, as it will be getting much more complex very shortly, but wanted a quick post to start things off even though I’ve already posted of some route redistribution from the OSPF section, such as redistributing connected routes.
Route redistribution is not distributing one routing protocol into another, it is taking a set of routes and putting them into another routing domain, including connected and static routes – This is a very important distinction. The Redistribution can be one way, as one side may only need to know the other sides routes, but for two-way communication to occur between routing domains you must either redistribute both ways OR static routes can be used to get there as well.
For redistribution, you must watch your entrance and exit points, as routing loops can occur easily if you have multiple routers sharing routes between domains (So for all intensive purposes the less ASBR routers the better).
A quick word on automatic Redistribution, it will ALWAYS involve EIGRP (at this time), such as the outdated IGRP (LAN Protocol) would automatically redistribute into an EIGRP routing domain if they shared the AS number. There are some other scenarios, however they involve EIGRP and AppleTalk / IPX, and I am going to leave it at that for now.
“redistribute …” in router config mode will always be how you begin the redistribution, and you will want to redistribute the desired protocols route in the current protocols configuration, which is incredibly hard to explain without this example:
To redistribute RIP routes into an ospf domain, you would go in the router ospf (process #) configuration mode and type in the following command:
“redistribute rip subnets” – Not a lot to it, like redistributing connected subnets
However to redistribute OSPF routes into RIP, you will need to add the metric to the routes you will be injecting on the ASBR, as RIP has a maximum cost / metric / HOP COUNT of 15 before the route is considered invalid. Being that this is the case, I used the same topology as posted in my last few OSPF posts, only the NBMA network was a RIPv2 domain and R1-R5 was my OSPF Area 0. This is the command issued on the ASBR in RIP config:
R1(config-router)#redistribute ospf 1 metric 2 <- For OSPF process 1, and the metric of 2
The metric is needed because both RIP and EIGRP do not have a ‘seed metric’ that they inherently assign redistributed routes (Like OSPF E2 routes), so it will bring in OSPF’s cost / metric into the RIP domain which will make the routes not appear at all because they are considered invalid due to their high cost / metric – and this is why it must be assigned.
***Very important to note, the metric set for OSPF routes being redistributed into RIP is the ‘beginning’ metric or hop count, which will then increment as it propagates through the RIP network – Be sure to keep that hop count below 15***
Another thing to note, which I found confusing, where as OSPF will label redistributed routes as E2 and I believe EIGRP is D EX, RIP routes learnt via redistribution still just have a route table code of R. So because the route has the R next to it, it may be learnt via route redistribution.
Another note, this is only so RIP can have other protocols redistributed into it, I will be doing a “wr er” on all routers now and finally making a new topology to look at changing Administrative Distances, and how to overcome suboptimal routing – more to come.