Re: [lisp] LISP Interworking: Proxy Egress Tunnel Routers

[Dino Farinacci <[e-mail]>]

> At least in theory, we get to replace several multi-homed sites,  
> each with their own PI address with a single PITR advertisement for  
> their EID block.
>
> Obviosuly, there is going to be some period of time when folks won't  
> trust the PITRs, etc...  But at least that is a theory which, if I  
> could believe the PITR deployment story, lead towards reduction.

Trust?

I don't have to trust all the routers that are getting this email  
message to you, but I do have to depend on them. On different for PTRs  
or PETRs.

Dino

>
> Yours,
> Joel
>
> Noel Chiappa wrote:
>>    > the PITR advertises in BGP a short prefix for a large block of  
>> EIDs.
>> Right, I forgot about that.
>> The problem with doing that is that if EIDs in that block are  
>> scattered,
>> topologically, you'll wind up with longer paths. TANSTAAFL...  
>> Course, that is
>> only true for traffic from non-LISP sites..
>>    > If it does not do this, then during transition (half of  
>> forever), we
>>    > get an increase in routing table size instead of a decrease.
>> Routing table size is a whole discussion in itself.
>> Even with a certain amount of reduction of splittage, you still  
>> wind up
>> needing to advertize both EIDs _and_ RLOCs, i.e. potentially more  
>> stuff. I'd
>> been hoping to be able to, before long, get to a situation where  
>> parts of the
>> network did not include all of (legacy-addresses, EIDs, RLOCs).
>> 	Noel
> _______________________________________________
> lisp mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/lisp