Re: [Sip] comments on draft-kaplan-sip-session-id-01

[Dale Worley <dworley at nortel.com>]

On Sat, 2009-03-21 at 11:28 -0400, Jonathan Rosenberg wrote:
> Now B and C are talking to each other. What is the session ID for this 
> dialog? B and C were both originally the UAS on different dialogs with 
> different session IDs!

The essential problem (which also shows up in
draft-loreto-sipping-context-id-requirements) is that you need to
express a correlation between two dialogs, but the correlation is
discovered after both of the dialogs have been initiated.

If I may be immodest, the References header could a solution:  State
that two dialogs are related to each other by (effectively) naming both
call-id's together, rather than by assigning them a common identifier
when they are created.

In regard to correlating the two legs of a call through an SBC, see
section 5.2 of draft-worley-references-02.  The SBC can put a References
header in a message going out on one leg that contains the call-id of
the dialog on the other leg, or vice-versa.  If the SBC wants to be
paranoid, it can create a pseudo-random pseudo-Call-Id and put it in a
References header in two messages, one going out each leg.  That's not
elegant, but it's effective and safe -- and doesn't depend on the UAs.

As far as I can tell, the References header handles all of the use cases
people have stated.

Dale