OSPF utilization with multiple exit points...?

[thek25]

I've taken it upon my self to elminate the multiple routing protocols and migrate to OSPF. Sounded really easy...Anyway digging deeper there are multiple exit points into the cloud. The exit points are not close to each other and there are no plans at eliminating them. I am almost thinking of putting BGP on the gateway links and running OSPF as the IGP and configure BGP to operate on the external links. Option 2, which a friend is leaning towards is raising the cost of the path on one of the routers between the multiple exit points. Would that eliminate loops? Or simply break the areas into smaller areas and have many ABR's that connect non-backbone areas. In my mind that would defeat the purpose of having redundant links. Mostly this is a mind exercise that has a slim possiblity, if my design is eally good, of implementation. Ideas or suggestions would be great. Size of the network exceeds 200 routers. Yes there are much smarter people than I that work here but they are happy with the status quo. I think it could be made better and operate smoother with less resources. Again, any help, suggestions or ideas would be cool.

 

[baddos]

The ip ospf cost works great. If you set it to a really high number, it will make it undesirable to use, and won't be installed in the routing table unless the preferred (normal) routes go away.

interface ethernet 0
ip ospf cost 100

 

[thek25]

Cool Thanks. After submitting the above I went and did more research. The cost idea seems to be the best and easiest approach. I am not very familar with BGP. Yeah I passed the 901 test but that does not make me a BGP guru. Just lets me know I fall into that category of just enough knowldege to really screw something up.

Thanks

 

[StarTAC]

well, u can simply use OSPF as your protocol to carry your infrastructure.. that is, loopback interface addresses... from experience, i recommend u carry customer prefixes using iBGP, and connect to your upstream carriers using eBGP...

iBGP is a bigger workhorse than OSPF.. and the smaller u can keep your IGP, the better for u.. that means, don't redistribute connected's, static's or even from BGP.. very very bad.. just use network statements to import routes from your loopback interfaces..

then put all your internal routes and customer routes into BGP..

to your upstream/Internet, announce your aggregated prefix..

good luck..

 

[thek25]

That is what I would really want to do, use BGP to connect higher. One thing I did leave out, kinda of important..is the network in question is a Government/military network. It hits 8 gateway routers before it hits the internet and I barely control the first. They are using EIGRP to control the links now and it is a mix of everything running internally. That is why I am mostly concerned with OSPF. I would love to run BGP between my gateway and higher that way my IGP is completely internal, backbone and all. It hits the ASBR and goes to soemthing else, no worries about any of the problems with OSPF and the backbone. What I want to be an ASBR now has to belong to area 0. Not what I would design if king for a day.

thanks

thek25
CCNA. 640-901