This may seem like a stupid question, but anyway:
When using dynamic routing protocols. Is it best practice to have the static routes distributed into them?
Is that the only way to really direct the traffic if not destined for 0.0.0.0?
IMO it really depends on the circumstances. If you have a specific scenario you're asking about, you might get a more definitive answer by going into more specifics in your question.
If I'm interpreting what you're asking correctly:
In a couple of production networks I worked on, there was no way to avoid redistributing static routes on our 'border router' for nets in other enterprises we connected to, who did not want to run a routing protocol on their border router. (We of course had a default route to the Internet). I know they ran a routing protocol within their own network, they just didn't want to deal with redistribution issues that occur when you run multiple routing protocols (and they had static routes on their side as well).
Thanks for the reply John.
I asking about a specific situation. I have a secnario that I want to simplify.
There are two differnet routing protocols on two different devices (switch and router connected by ethernet on the same subnet) they are both distributing their static routes into their respective routing protocols, then distributing mutually between themselvs. This all gets replicated out to the remote sites.
It is like you have no control over anything.
I want to remove some of it, but I can't take any sites down.
When you have dynamic routing protocols and if I did not have static redistribution, how would I direct traffic at the main site to different next hops, and make sure the remote sites are aware of it?
The only way to avoid the redistribution is to make sure all networks make it into the routing protocol.
This is only possible if all networks are directly connected to your (L3) switches and routers, and you can run a common routing protocol (layer two switches don't count in this). Then, you're set - design a routing architecture and go (and of course here's where the devil gets in the details - y'know, like how many routers / networks are we talking about, etc, etc.)
Do you think it is possible to add the static routes into the switch (the switch distributes static into RIP. The router then distributes RIP into OSPF) to the different subnets with out any problem?
Yes, that can work without any problem.
If I could, I would look at upgrading the switches' IOS to include OSPF, which would be better from an administration/configuration consistency standpoint.
I assume the switch with the static routes has VLANS and routing between them. Are the static routes needed because the VLANs are acting as transit networks (i.e. some device on the VLAN is routing to other networks but not running a routing protocol)?
I think that we need to be careful about saying that this would work without any problem. The last time that I tested this if you redistribute static into RIP and then redistribute RIP into OSPF then OSPF learns the RIP routes (the routes in the routing table with an R) but does not learn the statics (the routes in the routing table with an S). If you want OSPF to learn the static routes I believe that you need to redistribute static into OSPF.
The original question asked about best practice concerning redistribution of static routes into a dynamic routing protocol. I do not know that we can make generalized statements and say that it is best practice to redistribute (or best practice to not redistribute) static routes into the dynamic routing protocol. I think it depends on why the static route is there and whether other routers also need to know this. I have been in situations where static routes did need to be redistributed and have been in situations where the static routes did not need to be redistributed.
I was under the impression that RIP didn't mark static routes like OSPF (and others) does, so once they're being advertised, then as far as any other protocol was concerned there's no difference in the routes. I will dig into this in the lab - since you raise an exception, I need to double-check this.
I thought I was being pretty careful to say "it depends" and "it can", not "it will". To be truthful, there needs to be more specific information about the situation to give a good answer. Or maybe, the answer is, "no, it's not good practice, except unless there's no other way to get the job done". :o]
Hi .. in my experience it is not a good practice to redistribute in both ways as this could cause routing loops. I suggest to carefully identify what needs to be redistributed and if possible doing it on a one way only ..
You are correct that RIP does not differentiate external/redistributed routes from internal routes as most of the other routing protocols do. But in redistribution you learn a certain type of route. When OSPF is configured with redistribute rip, then OSPF will learn routes in the routing table that have an R. It is a subtle distinction but redistribute rip is saying to learn the routes in the table that RIP put there and is not saying to learn all the routes that RIP will advertise to its neighbors.
To look at it in a slightly different context: if one router redistributes static into RIP, and RIP advertises routes to a neighbor, it will advertise the routes that were originally static. And if the second router has learned those routes via RIP and the second router has OSPF with redistribute rip, then the OSPF will learn the originally static/now rip routes just fine. But if RIP has redistribute static and OSPF has redistribute rip on the same router then OSPF will not learn the static routes.
I do encourage you to take this to the lab and work through it. With 7 years experience as a certified instructor teaching Cisco courses, and in preparing for and passing the CCIE I have learned that there is no substitute for actually setting things up on real equipment and seeing what really happens.
And I do agree that if the question were a bit more specific or if it provided a bit more context then it would be easier to give a really good answer.
Thanks again Rick for the knowledgable answer.
Answers always seem to bring up more questions from me, but:
What would be the difference as far as the OSPF advertised routes that the remote sites saw?
Wouldn't the end result be the same as far as routing the traffic to the HQ switch thru the HQ router?
Unless the metrics were the difference.
There must be something in your question that I am not understanding when you say: "What would be the difference as far as the OSPF advertised routes..." I thought that it was fairly clear that if you redistribute static under router rip and redistribute rip under router ospf (but not redisribute staic under router ospf) that the result is that OSPF would not advertise the static routes. If you want OSPF to advertise the static routes (from the local router) then you must redistribute static under router ospf as well as redistribute rip.
Perhaps if you rephrase your question I might understand it better and be able to answer it.
Im sorry, it is me.
I saw your other post and did not read the last one carefully before I sent that last question.
I am pretty new to this forum and as as you can tell, I do not know as much as I probably should.
I am not clear or specific enough in my questions, and I get ahead of myself.
You are a great source of information and you all have been very patient.
For someone like myself, it is great help to communicate with people like you and the other guys in the forum.
I get frustrated when I have questions about my situation, and they usually are not answered easily by looking for the information.
So, I end up asking the question in the forum and sometimes it is not clear.
Please forgive initial failure to understand, it usually takes me several times to read the answers before it sinks in.
Sometimes it seems it is the simplest of things that I get confused about, for example, take a look at the floating static route post. If you are up for it.
Thanks for all of the valuabale information.