W3C home > Mailing lists > Public > public-ws-chor@w3.org > March 2004

Re: [wsbpel] SOA Patterns

From: David RR Webber <david@drrw.info>
Date: Fri, 5 Mar 2004 16:40:27 -0500
Message-ID: <075d01c402fa$79d5a2a0$6401a8c0@DRRW01>
To: <wsbpel@lists.oasis-open.org>, <public-ws-chor@w3.org>, "Paul Denning" <pauld@mitre.org>

Paul,

The answer to that is OASIS BPSS - where the pattern
behaviours for valid agreed interchanges are laid out
between two parties.

They then agree to follow those patterns.  BPSS is
adding some WSDL support - and essentially an
ASAP interaction within that BPSS is then just a
point node - where BPSS controls and manages
the pattern around that interaction.

You will find extensive work on UML and patterns
within the BPSS specifications and the CEFACT work
on business collaborations.

Thanks, DW.

----- Original Message ----- 
From: "Paul Denning" <pauld@mitre.org>
To: <wsbpel@lists.oasis-open.org>; <public-ws-chor@w3.org>
Sent: Friday, March 05, 2004 4:32 PM
Subject: [wsbpel] SOA Patterns


> I am wondering if W3C WS-Chor or OASIS BPEL TC are thinking about the
> following:
>
> Assuming minimal essential SOA includes services with descriptions (WSDL)
> and interaction via messages (SOAP), other things we might think of as SOA
> are just different ways of realizing SOA.  I call these "SOA Patterns".
>
> For example, some SOA realizations may have a BPEL engine for
orchestration.
>
> In some sense, the BPEL flow is just like any other web service with a
WSDL
> description and SOAP endpoint to invoke or start the flow.  However, SOA
> with BPEL seems special enough (in contrast to SOA without BPEL) that we
> might want to call it a SOA Pattern (and give it a name or URI to identify
> it like a namespace).
>
> Some realizations of SOA may not use UDDI for runtime discovery (late
> binding), while others depend on UDDI much like we typically depend on DNS
> to resolve hostnames to IP addresses.  SOA with UDDI may be another SOA
> Pattern.
>
> Some realizations of SOA may want to federate discovery.  There may be a
> BPEL flow defined that involves a query to one UDDI server (or a proxy) to
> find other registered (private) UDDI servers.  The BPEL flow would then
> query each of these UDDI servers for a specific service binding (e.g.,
> specifying a tModelBag).  The BPEL flow would join the results of these
> queries, and perhaps do some other consolidation or trimming.  It could
> even include a query to a semantic web match-maker before it queries UDDI;
> the semantic match helps construct the appropriate UDDI query.  If
> searching UDDI using a categoryBag and specifying a keyValue in a
> particular taxonomy, the BPEL flow may be able to invoke a thesaurus
> service to lookup similar terms in other taxonomies or ontologies.
Perhaps
> there is a private UDDI registry dedicated to a certain industry, like
> photography [1].  In the hypothetical BPEL flow above, where it looks for
> other UDDI registries, it could narrow the set of private registries to
> those that use a particular taxonomy (e.g., related to photography).  That
> is, I have a registry of registries where each registry can be assigned
> categories corresponding to the types of things that are in that registry.
>
> This BPEL flow may be exposed as a single service, which is itself part of
> another BPEL flow.
>
> Such an arrangement of services is clearly beyond minimal essential
> SOA.  It could be a SOA Pattern, or a way of implementing SOA within an
> organization or enterprise.
>
> If WS-CDL describes the interaction between organizations (e.g., B2B), and
> BPEL describes the flows within an organization (EAI or SOI), then how do
I
> describe the SOA Pattern?
>
> [1] http://production.pictureservices.org/directory/web
>
> Paul
>
>
>
> 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/wsbpel/members/leave_workgroup.php.
>
>
Received on Friday, 5 March 2004 16:40:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Saturday, 18 December 2010 01:00:54 GMT