[c-nsp] Converting OSPF backbone to iBGP

[Garry]

Mark Tinka wrote:
> On Tuesday 30 September 2008 22:59:34 Dan Armstrong wrote:
>
>   
>> In an "all iBGP" network - what do you do with customer's
>> that are using your address space?   Do you redistribute
>> Connected (& static routes) on your access routers into
>> iBGP?  I assume that's about all you can do, right?
>>     
>
> You can originate prefixes off the edge router in question 
> via the 'network' statement. This is what we prefer to do.
>   

Still trying to get this figured out ... take this setup e.g.:

    ROUTER1 ---- ROUTER2 ------ Backbone

ROUTER2 would be an iBGP capable router, connecting a POP site to the
backbone. ROUTER1 would be a dial-in router that nowadays hooks into the
WAN-wide OSPF area 0 ... originating IPs from both a local IP pool as
well as static IPs and networks assigned for dial backups and the likes
... ROUTER1 is not able to speak BGP of any kind, so the only choice
that remains is keep up the OSPF to ROUTER2 ...

But if I do that, I have all the routes it announces back in the
Backbone OSPF which I want to only carry the infrastructure IPs ... I've
already attempted to set up a second OSPF process in order to keep the
routes learned from one box out of the backbone area, but somehow I must
be missing something essential, as the router belonging to multiple
areas keeps mixing them together ...

So, any hint as to what I'm missing here? Or how I should go at it? Worst case (though not really _that_ bad, as there are only a very limited number of routes/prefixes affected) would be I just keep the OSPF-only stuff in my backbone Area0, while moving everything else into iBGP ... not a very clean/consistent setup, but would work none the less ...


Tnx!