Right - another example is asynchronous messaging queuing, like what we're doing with AMQP, which is becoming more and more important as SOA tends more and more toward loosely-coupled interactions. In total the industry could really benefit from some better "top down" or "contract first" tooling that defines the service contract first and then allows a choice of implementation. If you tie the tools to OO, and derive the service from an object, it's more difficult for the SOA to encompass the variety of technologies and "orientations" that exist.
Eric
----- Original Message ----
From: Steve Jones <[EMAIL PROTECTED]>
To: [email protected]
Sent: Sunday, August 20, 2006 1:04:25 PM
Subject: Re: MDA/UML/OO and SOA (was Re: [service-orientated-architecture] John on Gartner, AJAX & Assorted TLAs)
From: Steve Jones <[EMAIL PROTECTED]>
To: [email protected]
Sent: Sunday, August 20, 2006 1:04:25 PM
Subject: Re: MDA/UML/OO and SOA (was Re: [service-orientated-architecture] John on Gartner, AJAX & Assorted TLAs)
One issue I've got with UML for SOA modelling is that its fine when thinking about software, but a bit rubbish when thinking about the business. Its also a little "rich" in terms of communication to the business. Another challenge is that some services are best implemented using BPM, or Human Workflow, type tools which really don't fit into the UML language.
UML is fine for some SOA systems, and a bit rubbish for others, what we do need (IMO) is something that sits above all of these different approaches and models just the services and their interactions (not process) then we can pick the right implementation modelling technology for each of the services.
Steve
On 20/08/06, Stefan Tilkov <stefan.tilkov@ innoq.com> wrote:
On Aug 19, 2006, at 2:31 PM, Lukas Barton wrote:
> Stefan Tilkov wrote:
>> +1. In fact, I don't think there has ever been a "good" UML tool -
>> in the sense of "good for automated software development".
> We could start DML vs DSL debate ;-)Not much of a difference to me -- DML graphical, DSL textual,
otherwise similar enough so it doesn't matter.
> Do you agree that: "Proprietary tools which worked mostly were DSL."?
>No. In my personal experience, most model-driven projects, SOA or
not, were driven from UML models, more or less heavily customized via
UML profiles.
>> A UML profile is a poor man's metamodel, so this could
>> conceptually be used with MOF or EMF, too. The problem is that one
>> has to define what the exact SOA metamodel is :-)
> But there is no standardised profile for SOA.
> OMG started SOA SIG (http://soa.omg. org/) only few month ago. Can
> you guess whether this group goals could be successful? (eg. create
> SOA metamodel, ...).
> May be I overestimate necessity of standards. But without standard
> you get a mess (eg. not cooperating tools...).
>Even with a custom UML profile, the commonality is large enough for
people to understand your models because UML is so widespread. If
only the tools sucked less ...
Stefan
--
Stefan Tilkov, http://www.innoq. com/blog/ st/
SPONSORED LINKS
| Computer software program | Computer software spy | Computer job |
| Database software | Discount computer software |
YAHOO! GROUPS LINKS
- Visit your group "service-orientated-architecture" on the web.
- To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]
- Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
__,_._,___
