W3C home > Mailing lists > Public > www-ws-arch@w3.org > May 2003

Re: Web Service Description and stateful services

From: Christopher B Ferris <chrisfer@us.ibm.com>
Date: Tue, 20 May 2003 07:43:25 -0400
To: Sergio Andreozzi <sergio.andreozzi@cnaf.infn.it>
Cc: www-ws-arch@w3.org
Message-ID: <OF34535299.662F494A-ON85256D2C.003D46E1-85256D2C.00406657@us.ibm.com>

Sergio Andreozzi wrote on 05/20/2003 07:03:16 AM:

> 
> Dear WS-ARCH group members,
> 
> the Web Services description is currently defined in the latest Web 
> Service Architecture document (1.5.3) as:
> 
> "The service description is a machine processable specification of the 
> message formats, datatypes and protocols that should be used between the 

> requester agent and the provider agent. It also specifies the network 
> location of the provider agent, and may provide some information about 
> the message exchange pattern that is expected"
> 
> In the research community, I've seen some attempt at representing the 
> service state by using WSDL extensions, i.e. by extending the web 
> service description with attributes that carry the information related 
> to the current state of a service. This means that, for each service 
> instance, a WSDL document should be created.

Yes. And, I assume by "research community" that you refer to the
work on the GRID? Or is this some other similar work?

> 
> I would like to receive your point of view on the following issue. From 
> the definition you give, I understand that the service state is a 
> concept more related to the service representation (def. 2.2.29), while 
> the service description (def. 2.2.32) should describe service interface 
> and semantics.
> 
> Extending a WSDL document with service representation related data is 
> somewhat merging service description with service representation. This 
> solution might be handy when you want, for instance, advertise the 
> quality related attributes in order to make a QoS based discovery.

Agreed, it is blurring the distinction between the two concepts, but as 
you 
point out, it can indeed be useful for such things as QoS-based discovery.
Of course, you could also make use of the Notification service and
have instances advertise their service data in that manner and still
be able to effect QoS-based discovery (assuming that some discovery
agent is facilitating the job by subscribing to servicedata, etc.)

> 
>  From a Web Service Architecture point of view, how do you deal with 
> this issue and the above solution?

Well, this really should be discussed ay length in the WSD WG as part
of the process of incorporating such an extension in WSDL1.2.

I think you raise an interesting point, and that you should do so in the
context of WSD WG to call it to their attention.

Cheers,

Chris
Received on Tuesday, 20 May 2003 07:44:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:19 GMT