Redistributing routing protocol question

[limtohsoon]

Hi Sir,

Please see attached network diagram.

The center router is running 3 routing protocols: OSPF process 1, OSPF process 2, and RIPv2.

I'd like to perform mutual redistribution among these 3 protocols. Do I need to use route-maps to set tags and then redistribute based on the tags, to prevent potential route feedback? Is there a possibility of routing loop in this topology?

Please advise.

Thank you.

B.Rgds,

Lim TS

[rais]

You can use route-maps with redistribute to advertise routes from one process/protocol to the other.

If all OSPF needs to talk to all other OSPF, may be OSPF 2 can be removed with all inferaces placed in a single OSPF process.

Hope this helps.

[Edison Ortiz]

You have a single point of route-redistribution, therefore there won't be any route feedback from any other routing protocol.

You can use the route-map concept for better control of the routes or a simply redistribute [protocol] would do. Don't forget to include the metric from each protocol when performing the redistribution.

Good Luck.

[limtohsoon]

Hi,

Thanks for your response. I also believe routing loop won't happen here because there's only a single boundary router performing the mutual redistribution.

Some other questions:

(1) When redistributing into OSPF, I usually do not specify a default metric, letting it default to 20. Is this best practice? However I do control the metric-type, i.e. E1 or E2.

(2) When redistributing OSPF into other routing domain, I usually include "match int ext 1 ext 2". Without these keywords, what are the default matches? Matching OSPF internal routes only?

Thank you.

B.Rgds,

Lim TS

[royalblues]

Lim,

It is always a good practice to define the metric when redistributing into OSPF. The default of 20 is the seed metric applied when no metric is defined exclusively except for BGP in which casde it is 1

Metric type can be controlled during redisribution in the following way

redistribute BGP 1 metric 10 metric-type

If you configure the redistribution of OSPF into BGP without keywords, by only OSPF intra-area and inter-area routes are redistributed into BGP, by default

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a00800943c5.shtml

HTH

Narayan

[Edison Ortiz]

RESPONSE INLINE:

____________________________________________

Some other questions:

(1) When redistributing into OSPF, I usually do not specify a default metric, letting it default to 20. Is this best practice? However I do control the metric-type, i.e. E1 or E2.

_____________________________________________

OSPF is not such a problem since it defaults to 20. The only thing with external routes into OSPF is to include the 'subnets' keyword.

As for external routes into RIP, if you don't specify the metric, the routes will not be injected.

What you are referring above, are metric-type which is different that my previous suggestion. By default, all external routes are E2 - which means the metric won't be changed on a hop-by-hop basis.

+

______________________________________________

(2) When redistributing OSPF into other routing domain, I usually include "match int ext 1 ext 2". Without these keywords, what are the default matches? Matching OSPF internal routes only?

______________________________________________

Without those keywords, all OSPF routes in the RIB will be redistributed into that routing protocol.