draft-rosen-l3vpn-mvpn-bidir-01.txt

[Eric Rosen <[e-mail]>]

I would like to call the WG's attention to draft-rosen-l3vpn-mvpn-bidir-
01.txt.

This draft provides the specification for the use of bidirectional tunnels
to instantiate I-PMSIs or S-PMSIs.  The bidirectional tunnels can be created
by using BIDIR-PIM (already standardized) or by using mLDP (already adopted
by the MPLS WG).

The use of bidirectional P-tunnels is discussed in draft-ietf-l3vpn-2547bis-
mcast-10.txt in numerous places.  In draft-ietf-l3vpn-2547bis-mcast-bgp-08,
the means of using the PMSI Tunnel Attribute in BGP MCAST-VPN routes to
identify bidirectional P-tunnels has already been standardized.

Draft-ietf-l3vpn-2547bis-mcast-10.txt specifies a method of supporting
customer BIDIR, known as "partitioned sets of PEs".  This method is also
recommended by draft-ietf-l3vpn-mvpn-considerations-06.txt.  The use of
MP2MP LSPs is one of two possible methods of implementing "partitioned sets
of PEs", and is the only method of implementing this recommended method when
upstream-assigned MPLS labels are not available.

Given the above, the detailed specification of the use of bidirectional
P-tunnels is an obvious "next step" for the WG to take on.  Most of the
material in this draft has been available for several years, and as of today
I am not aware of any competing drafts.

Note that in the -01 version of this draft, it is no longer always required
for the root of an MP2MP LSP P-tunnel to be a PE router.  This removes (I
hope) one of the objections that was made in Anaheim during a brief
discussion of this topic.

I regret that I will not be in Maastricht to discuss this in person, but
comments on the mailing list are welcome.