RE: Addressing Orchestration

Mark,

While I do not disagree with anything you say below, the real
orchestration issue (IMHO) is about knitting together multiple web
services. What you describe below relates to enhancing the description
of using an individual web service. This certainly has a role, but a new
workgroup should 
address the spectrum.

Martin.

-----Original Message-----
From: www-ws-arch-request@w3.org [mailto:www-ws-arch-request@w3.org] On
Behalf Of Mark Hapner
Sent: Monday, June 24, 2002 10:20 PM
To: www-ws-arch@w3.org
Cc: 'www-ws-arch@w3.org'
Subject: Re: Addressing Orchestration




I agree with Ugo that the Web Services Architecture working group
ideally should be working up a conceptual architecture that could form
the framework from which individual work efforts relating to web
services will be launched. 

However, I also agree with Dave Orchard that this is a daunting task for
the Web Services Architecture working group, and that they already have
a lot on their plate. Therefore, I think that the best forward is to
establish a Web Services Choreography working group, as it has already
been discussed, and use that working group to seed the work on the next
level up of the architecture. 

The currently existing Web Services Description working group is in
essence concerned with describing a set of web service operations, or
stated differently a static web service. This is an  absolutely
necessary foundation for anything more complex that one might want to do
with web services. However, we are now ready to move beyond static web
services, into stateful web services that may have complex usage
scenarios. 

It is this space that a choreography working group should be addressing.


How does one define the supported usage of a stateful web service?  In
essence this means defining a number of relationships between individual
operations. 

What are the allowable sequences of usage of those operations?  

What are the relationships (correlations) between the data exchanged? 

What are the transactional semantics (if any) of a sequence of
operations? 

How would one compensate for a sequence of operations, since atomic
roll-back is not relevant. 

What are the exceptions that may prevent a sequence from completing, and
how are those exceptions handled? 

All of these aspects of a stateful web service need to be formalized and
described in an unambiguous and coherent syntax in order for web
services to be adopted for anything beyond the most atomic usages. 

This is the challenge of the next layer of the web service stack, on top
of WSDL, and the web services architecture working group should start
meeting this challenge, either directly, or indirectly by recommending a
new choreography working group.

-- Mark


> Ugo Corda wrote:
> 
> Hi everyone,
> 
> Since I just joined the WSA WG, I thought of warming up for the task 
> by asking a question prompted by the recent publication of the WSCI 
> spec(http://wwws.sun.com/software/xml/developers/wsci/).
> 
> It looks like Orchestration/Choreography/Workflow proposals relevant 
> to Web Services are proliferating in the industry. The ones that come 
> to my mind are IBM's WSFL, Microsoft's XLANG, BPMI's BPML, ebXML's 
> BPSS, HP's WSCL, and now WSCI.
> 
> My hope is that clarity will soon be made in this important area(s) of

> Web Services utilization. As usual, lack of clarity increases the 
> risks of slow adoption of, or resistance to, Web Services technologies

> in many parts of the industry.
> 
> I wonder if WSA could play a role in this area, at least from the 
> point of view of establishing a conceptual and architectural framework

> within which the various proposals can be positioned, discussed, 
> compared and selected. (Please understand that I am not talking about 
> discussing specific proposals and/or establishing profiles, a role 
> which is already being played by other organizations, e.g. WS-I).
> 
> I looked at the Requirements document and it does not seem to mention 
> Orchestration or use any similar word. Browsing through the archives 
> of this WG, I see that the subject of Orchestration has been raised a 
> few times in the past, but my impression is that no final decision was

> reached regarding whether or how to address it. If that is the case, 
> the recent publication of yet another Orchestration spec could be the 
> opportunity for the WG to put a stake in the ground in this particular

> area.
> 
> Regards,
> Ugo
> 
> Dr. Ugo Corda
> Standards and Product Strategies
> SeeBeyond Technology Corporation
> 404 E. Huntington Dr., Monrovia, CA 91016
> (626) 471-6045 -- phone
> (626) 353-4851 -- cell
> (626) 471-6021 -- fax

Received on Tuesday, 25 June 2002 18:38:07 UTC