Re: draft-ietf-mpls-rsvp-te-p2mp-06.txt [P2MP ID]

[Lou Berger <lberger@labn.net>]

Yakov,
        Thank you for the concise response to a question that has 
been outstanding for some time.  Now there are two specific aspects 
to the proposed response (Rahul's) to the issues you raise below:

1) Tunnel ID
[From rahul's mail]
 > 4.
 > 19.1.1
 >
 > "Tunnel ID
 >
 >       A 16-bit identifier used in the SESSION object that remains
 >       constant over the life of the P2MP tunnel."
 >
 > to
 >
 > "Tunnel ID
 >
 >       A 16-bit identifier used in the SESSION object that remains
 >       constant over the life of the P2MP tunnel. It SHOULD be set to 0
 >       by the ingress LSR and be ignored on receipt."
 >

I don't see how this change, setting the ID to zero, relates to the 
issue you raise.  Can  justify why this change is required?

2) Extended Tunnel ID
[again from rahul's mail]

 > 5.
 >
 > "Extended Tunnel ID
 >
 >  A 32-bit identifier used in the SESSION object that remains
 >       constant over the life of the P2MP tunnel.  Normally set to
 >       all zeros. Ingress nodes that wish to narrow the scope of a
 >       SESSION to the ingress-PID pair may place their IPv4 address
 >       here as a globally unique identifier [RFC3209]."
 >
 > to
 >
 > "Extended Tunnel ID
 >
 >       A 32-bit identifier used in the SESSION object that remains
 >       constant over the life of the P2MP tunnel. This identifier
 >       MUST be set to the ingress LSR's IPv4 address."
 >

So the original text, allowed for global uniqueness and included a 
"well-established procedures for assigning (globally) unique" P2MP IDs.

It seems to me that the -05 text covered the issue you raised.  So 
now we come to the heart of the matter:  Why is a change in the -05 
definition of the IDs needed?

It seems to me that at most the text needs to replace a "may" with a 
"MUST", as in "Ingress nodes that wish to narrow the scope of a 
SESSION to the ingress-PID pair MUST place their..."

Lou

At 10:43 AM 6/27/2006, Yakov Rekhter wrote:

> Lou,
>
> [since the answer to the question you posed below may benefit
> other folks, I added mpls and ccamp mailing list to the cc:]

a bit unusual to redirect private mail, but whatever...

> > Rahul,
> >          Okay, so I'm just a bit slow.  Can you explain why the
> > semantics of P2MP-TE needs to be different than P2P-TE?
> >
> > Yakov,
> >
> > Perhaps you should answer this?
> >
> > Lou
>
> To answer your question observe that in P2P-TE the Session object
> carries IPv4 tunnel end point address, while in P2MP-TE the Session
> object carried P2MP-ID.
>
> Here are some of the reason(s) semantics of IPv4 tunnel end point
> address needs to be different than P2MP-ID:
>
> 1. IPv4 tunnel end point address is a unicast address. P2MP-ID
> is not.
>
> 2. IPv4 tunnel end point address is suitable for hierarchical routing
> (e.g., CIDR). P2MP-ID is not.
>
> 3. There are well-established procedures for assigining (globally)
> unique unicast IP addresses. There are no such procedures for
> assigining (globally) unique P2MP IDs.
>
> Yakov.