[c-nsp] Unified MPLS - Discrete area or separate IGP in Access Layer

[Mark Tinka]

On Thursday, February 06, 2014 04:36:45 PM Adam Vitkovsky 
wrote:

> This is regarding Unified MPLS for LTE aka Hierarchical
> MPLS with RFC3107. I'd like to know whether there are
> networks out there running separate IGP processes per
> each access or aggregation network. Or whether you are
> running common IGP and using separate area/level for
> each aggregation/access network please?
> 
> I'm interested for any pros/cons either of these
> solutions have. I guess it's no biggie whether there are
> going to be separate processes in access and core not
> talking to each other or just separate areas. Thank you
> very much.

There has been quite a bit of talk about this in past posts 
on this list.

Generally, I'd recommend IS-IS because as a single Level 
IGP, it scales very well. You also want a single Level 
because some MPLS-TE features don't work well across Levels 
(although I know those restrictions are being lifted with 
SR, but that's still a while away).

What Saku, I and others have loosely agreed on for the way 
forward for deploying large scale IP/MPLS networks, edge-to-
edge, is:

	- IS-IS L2-only in the core
        - IS-IS L2-only in the edge
        - IS-IS L2-only in the Metro-E Access
        - BGP-LS between Metro-E rings and edge
        - SR within IGP domains

Some of these may not be feasible now (like BGP-LS and SR), 
but you could start with a base that leads you to that, and 
I think RFC 3107 can help there.

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20140206/df404853/attachment.sig>