[j-nsp] redistributing label between rsvp and ldp

[Mark Tinka]

On Monday, April 30, 2012 04:00:09 PM Keegan Holley wrote:

> I assumed you were exploring the configuration of the
> signaling protocols. My point was that I can't think of
> a situation (not saying that one doesn't exist) where I
> would run both protocols on purpose. At most it would
> happen during a cutover from one to the other.  I can't
> think of many things one protocol does that the other
> doesn't that are important enough to keep both around
> permanently.

Until recently, p2mp LSP's were only available on RSVP-TE. 
Now, mLDP also supports p2mp LSP's (and has room for native 
mp2mp LSP's in the future).

For our IPTv deployment, we had RSVP-TE providing p2mp LSP's 
re: the NG-MVPN architecture, and other MPLS traffic being 
supported by p2p LDP (including LDPoRSVP LSP's).

Since mLDP now provides p2mp LSP's, and can be used to 
signal the data planes that Multicast traffic will follow, 
one may do away with RSVP-TE altogether. But there is still 
a reason to retain RSVP-TE, and that is FRR.

Our thinking was:

	o Use RSVP-TE for IPTv p2mp LSP's, as it provides
	  50ms protection.

	o Use mLDP for data-based Multicast, as it scales
	  better.

Mark.