A piece we recently ran on this site stared a minor to-do on the estimable Yahoo Groups Service-Orientated-Architecture...
thread. The SearchSOA.com story was called ''Gartner AADI Summit: SOA going into 2009'', but, as it ricocheted back and forth like a bleeping cursor in the old electronic pong game, it came to be known on the thread as "Yefim Natis is sure 'SOA is Integration.'" Why?
Well, during the course of the SearchSOA story, Gartner's Yefim Natis was quoted as saying 'SOA is integration.' This didn't sit right with some people, who thought it was at best, a return to notions of the old days of EAI or, at worst, the equivalent of a hex that was going to upset the holy apple cart of modern SOA. Asked one: How can we end this madness?
First: In fairness at least to Yefim Natis, I would like to point out that he described other aspects of SOA and these were noted in the story, and that, naturally, a brief story is just a snapshot that does not include each and every aspect of a conversation, and, yes, Natis even described other aspects of SOA that were not included in our item.
Second: What is wrong with SOA as integration? That is what application development has really been about for a long time. If we focus so much on architecture that we find it hard to admit that integration remains the daily mandate, our view may become obscured. Yes, with an emphasis on standards and services, SOA is delivering greater flexibility than EAI did in days of yore. Yes, it is important to root out badly architected legacy applications, and not paper over them. But we have to regularly deliver, and that usually means delivering better integrations of applications. You must plan, yes, but at some point it is 'time to make the donuts'.
Third: I think Yefim Natis puts it better: "The difficulty in SOA-oriented development is that it must achieve real short term business goals while setting the stage for far-reaching architectural objectives." There is a trick to modern development, and its not a trick everyone can turn. Balancing delivery today against a plan for tomorrow is the key.
Funny thing though, "AADI Summit: SOA going into 2009" did suggest that a consensus was arising as to what SOA was. That and other assertions may be questioned. Hopefully, despite differing impressions, differing emphasis, and different analysis of details, a SOA consensus is emerging.