Nice one Jean Jacques, I totally agree with you! Everybody is 'doing' SOA, but I don't see a lot result architecture's conforming to some basic principles you tend to expect from a SOA.
Uhh - I didn't deliver anything personally. I chaired the OASIS SOA RM Technical Committee, currently and historically one of OASIS's largest ever TC's. Over 225 individuals were involved in the TC and over 18 months it delivered an OASIS standard that had the highest number of "yes" votes anyone can remember. It is also heavily and successfully used in many verticals.
I am very proud of the model and think the people who worked on it were in fact very competent and did extremely good work. If you want to be constructive, get involved with the TC and tell them specifically what you don't like about it. Don't cower in the corner and speak behind people's backs.
You are right in asserting that many people still don't "get" SOA - that was the whole point of proposing a Reference Model - to have a template and tool for discussing building specific systems around agreed core concepts - trashing the OASIS SOA reference Model is simple and cheap but, honestly - do you have anything better to propose? And that is sufficiently technology- and implementation-independent that SOA doesn't gethijacked by a particular vendor or technology stack? That was the OASIS initiative's challnge, and I think it did what it set out to do
Duane:
yes, I stand corrected, you don't deliver much in general, you just take other people's ideas, and we can see what you do with them.
Peter:
I don't put myself in the category of people who cheaply comment on things. I wrote a whole book on my vision for SOA and it can be downloaded here: http://www.infoq.com/minibooks/c...re-
construction . I don't think there is anything vendor specific in this book. I am completely vendor neutral. I don't work for a vendor.
The reason why I think the SOA RM is a piece of crap and counter productive is that it does not lead to any practical programming model to build service oriented solutions. At best it is a "Service RM", it is certainly NOT a SOA RM. It does not even have the concept of "coordinator" in it. There is no emphasize on peer-to-peer relationship. There is no resource concept. You talk about a process model, but there is no notion of role. There is no notion of user either. Yeah, such a great piece of work.
Come on, Peter, how can you be proud of something like this? It is no surprise that hardly anyone is using it.
This is the most synthetic view of a SOA reference model I came up with (I am not claiming it is complete by any means):
http://www.ebpml.org/blog/48.htm
Commenting by HaloScan
