On Fri, Dec 18, 2009 at 9:14 AM, M. Ranganathan <[email protected]> wrote:
> Hello,
>
> I am building a sip tester where I determine relatedness ( or more
> precisely, causality)  between a sent and received message so that the
> tester can only simulate activity between endpoints of interest. I can
> use Via for determining causality between sent and received endpoints
> when there are no intervening B2BUA. Unfortunately, that this does not
> work when you have an intervening B2BUA.
>
> To see this, consider a UAC sending an INVITE through a B2BUA that
> re-originates the INVITE using a different Call-ID.

And further it does not preserve the Via headers of the inbound Client
Transaction as it re-originates the request.


 How do you
> connect the UAC (Client Transaction) that sent the request with the
> UAS (Server Transaction) that got and processed the request?
>
> For this, the References header comes in handy. However, there is
> still a problem in that it does not disambiguate between two fork
> targets to the same endpoint ( they will both have the same call ID).
> Second,  call-ID alone is not enough to distinguish between two such
> B2BUA  originated requests in close succession. It says nothing about
> ordering if these belong to different Dialogs on on the UAS. For now,
> this is not a problem because we do not have such a case. Call-ID
> alone is enough to determine the server transaction of interest.
> However, to make it truly general, I think we need to add a tag
> parameter to the References header so that the B2BUA can compute such
> a tag to relate a Server Transaction to a Client Transaction that was
> causally connected to the inbound Server Transaction.
>
> Incidentally, here is another excellent reason to implement the
> References draft -- it allows one to build testable components in the
> presence of such nasties as B2BUAs.
>
>
> Do we need such a tag?
>
>
> Regards
>
> Ranga
> --
> M. Ranganathan
>



-- 
M. Ranganathan
_______________________________________________
sipx-dev mailing list [email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-dev
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-dev
sipXecs IP PBX -- http://www.sipfoundry.org/

Reply via email to