Re: [mpls] Regarding draft-wijnands-mpls-mldp-csc-02.txt

[Eric Rosen <erosen at cisco.com>]

John> contrast to the existing CsC solutions if possible

That's a fair request.

The solution described in sections 15, 11.1.2 and 11.3.2 of
draft-ietf-l3vpn-2547bis-mcast-bgp presupposes the use of mLDP on the
interface between the "outer carrier" and the "inner carrier".  It uses BGP
C-multicast routing to convey the outer carrier's mLDP signaling
information, and uses BGP S-PMSI A-D routes to assign the outer carrier's
mLDP-based P2MP LSPs to P-tunnels of the inner carrier.  If a particular
deployment supports upstream-assigned MPLS labels, this can be used to allow
aggregation of multiple P2MP LSPs from the outer carrier into a single
P-tunnel of the inner carrier.

Now consider a deployment scenario with the following properties:

- The inner carrier has not deployed and/or does not want to deploy BGP
  C-multicast routing.

- The inner carrier and the outer carrier both use mLDP for their P-tunnels.

- The inner carrier is not concerned about aggregation (either because
  upstream-assigned labels are not available, or because it doesn't want the
  sub-optimal routing that comes with aggregation).

In this scenario, the simple "in-band" signaling of draft-wijnands-mpls-
mldp-csc provides a much simpler solution.  We think this will be a quite
common deployment scenario.  

Of course, if a carrier's carrier does want to do aggregation and does want
to deploy BGP C-multicast routing, those options are certainly available in
the overall solution set.