RE: [wsdm] Long-lived interactions requirement proposal

["VAMBENEPE,WILLIAM (HP-Cupertino,ex1)" <vbp@hp.com>]

Hi Shel,

You are right that the discussion you refer to below is an important one and
impacts our work, but I think it impacts us in terms of deciding how much we
rely on OGSI for our infrastructure (the MUWS part) but doesn't really
impact the question of how we address management of services engaged in
long-lived interactions in the MOWS part of our work. The reason why I
believe this is the case is that the requirements that I propose do not make
any assumption on the correlation model. All it assumes is that a web
services sends/receives messages and somehow knows that these messages are
related. Whether this happens through the mechanisms used by BPEL/WSCI,
through the mechanisms provided by OGSI or by some other mechanism (for
example at the transport level) is irrelevant at this stage. All these
approaches to correlation could be managed the same way if we accept the
requirements proposed in my email.

Where it would start to fork is if we went further and decided to provide a
management interfaces for services that obey a particular conversation
description language or that specifically use Grid mechanisms for
correlation. But this is work that is out of scope for our initial document.
What we are doing right now is generic enough that it is common to any
correlation model as long as it is based on the idea of messages being
exchanged which I believe can be taken a given for web services.

BTW, the WSMF Conversation managed object which is one possible way to
implement the requirements I propose was originally designed with BPEL-type
interactions in mind. So this is clearly supported. Because it is generic
enough it happens to also nicely cover management of Grid-like long-lived
interactions.

Another thing your message point out is that the word "session" might not be
the best one either. I picked it in this proposal because it seemed to be
the one that people were most comfortable with during Thursday conf call.
When we came up with WSDL, we had numerous discussions about how to call
this entity, "conversation", "session", "context", "interaction", etc... I
am happy with any of these as long as we clearly define it as one service's
view of a set of related messages.

Regards,

William


> -----Original Message-----
> From: Shel Finkelstein [mailto:shel.finkelstein@sun.com] 
> Sent: Monday, September 08, 2003 9:14 AM
> To: 'wsdm@lists.oasis-open.org'
> Cc: VAMBENEPE,WILLIAM (HP-Cupertino,ex1)
> Subject: Re: [wsdm] Long-lived interactions requirement proposal
> 
> 
> Sorry that I was not able to participate in Thursday's call.
> 
> Serious questions have been raised from several directions 
> (not just in the
> Newcastle/Arjuna paper) as to how the work in BPEL4WS/OASIS 
> WS-Business
> Process group and WSCI/W3C WS-Choreography group relate to 
> the GGF OGSI
> spec.  WSDM will discuss this later this month, when Steve 
> Tuecke returns
> from vacation.  Issues include differences in programming 
> model between
> message correlation in BPEL/WSCI and locators in OGSI, including
> scalability, distributed systems program model 
> (instance/session vs process
> model), uniformity (two models vs one) and flexibility.
> 
> My question is:  Are we writing requirements assuming a 
> session model before
> we've had the (postponed) discussion as to whether that's the 
> consensus
> model?  My own perspective is that the WSCI/BPEL model (also 
> used in JSR
> 207-208) is superior, and the industry should converge on that model.
> 
> I've expressed this opinion to the GGF OGSI group (superior 
> model), but
> regretably my understanding of the issue came too late to 
> have any impact on
> this iteration of the OGSI spec.  I'd rather not see this 
> recur within the
> WSDM group.
> 
> Regards,
> 
> Shel
> 
> 
> "VAMBENEPE,WILLIAM (HP-Cupertino,ex1)" wrote:
> > 
> > Hi all,
> > 
> > As agreed during the conf call on Thursday, here is a 
> proposal on how to add
> > support for services involved in long-lived interactions to our MOWS
> > requirement doc.
> > 
> > Regards,
> > 
> > William
> > 
> > Proposal:
> > ---------
> > 
> > In the glossary: Remove all references to conversations and 
> define a session
> > as "a set of related messages (sent or received by the 
> service) and their
> > context".
> > 
> > In the requirements list (section 2), add:
> > 
> > - The manageability representation MUST allow monitoring 
> (including state
> > and access to messages) and control of the current sessions 
> of a service.
> > - The manageability representation MUST allow metrics to be 
> defined at the
> > session level.
> > 
> > To unsubscribe from this mailing list (and be removed from 
> the roster of the OASIS TC), go to 
> http://www.oasis-open.org/apps/org/workgroup/wsdm/members/leav
e_workgroup.php.