[c-nsp] IP LFA in ring topology

Phil Mayers p.mayers at imperial.ac.uk
Fri Nov 30 05:35:08 EST 2012


On 11/30/2012 08:46 AM, Adam Vitkovsky wrote:
>> global labels would even allow us to get rid of LDP, by distributing
> labels via ISIS
> And that would be AWESOME, I didn't quite get the idea of creating other RIP
> like protocols in favor of existing link-state protocols -same goes for PIM
> vs ISIS

You mean like MOSPF?

Frankly, I don't see why you'd want to load your IGP up with the soft 
state from multicast; it can be potentially large, and 
frequently-changing. I *like* that PIM runs above the IGP.

Also, PIM has to deal with the whole register/register-stop and other 
junk. So it's not just carrying routes.

LDP is of course a slightly different matter. It may indeed have made 
sense to put labels in the IGP rather than LDP. But there are things LDP 
can do (e.g. multi-hop sessions for PW) that IGP can't. And once you've 
got LDP for those, you may as well leave the IGP alone - I guess that 
was the reasoning.


More information about the cisco-nsp mailing list