RE: Proposed text on 'SOA' (resend)

 

> -----Original Message-----
> From: Jim Webber [mailto:jim.webber@arjuna.com] 
> Sent: Wednesday, September 10, 2003 7:00 AM
> To: 'He, Hao'; 'Savas Parastatidis'
> Cc: www-ws-arch@w3.org
> Subject: RE: Proposed text on 'SOA' (resend)


We're definitely getting to something that should be in the Arch document.
Do we have a concept of "basic stateless web services" that implement one
message exchange pattern, and "composite web services" that consist of
multiple MEPs that are associated/correlated/choreographed using WS-Context,
WS-Transaction, WS-Coordination, WS-CHOREOGRAPHY, etc., and which maintain
their state via a number of mechanisms? (I'm sure the definitions and
examples here are muddled, but bear with me.)
 
> This I think is Savas' key 
> simplification - Web Services exchange messages, and anything 
> over and above that assertion is dangerously close to being 
> at the application level.
> 
> This also seems to dovetail nicely with mechanisms for state 
> management (eg, WS-Context, WS-Coordination) which are 
> deployed to support particular applications.


I think I agree, except for the "dangerously close to being at the
application level. It seems to me that the whole point of these "mechanisms
for state management" is to provide a layer of service that is between the
basic web services and the application.  Of course, the application CAN do
all this (and RESTifarians generally argue that applications SHOULD do it),
but the point of these "dovetailing" efforts is to shove this down into the
infrastructure.

Received on Wednesday, 10 September 2003 10:13:51 UTC