IMO, Kumar is right and Joe is wrong, that simple. Legacy apps in IT never been services (if they were not designed as such) and they are not services (under the same conditions). The best thing the Legacy apps can do for SOS is to become reliable RESOURCES. The "“modern” systems" operation based on SO principles not wrap but shield Legacy apps when treat them as resources.
This is why I think that so-called 'Composite Applications' so dear to Sun Microsystems have very little to do with SOA. There 'Composite Applications' are classical integration products with no service orientation in them. If the goal is 'reuse legacy systems', then do it, and SOS is not needed to reach this goal. - Michael ________________________________ From: Gervas Douglas <[email protected]> To: [email protected] Sent: Thursday, March 19, 2009 3:41:37 PM Subject: [service-orientated-architecture] Joe on SOA without service-enabled apps <<“SOA Possible Even Without Service-Enabled Apps.” This is a statement that goes against the conventional wisdom, so, being a fan of things that go against conventional wisdom, I checked out this Q&A interview with Shailender Kumar, vice president of Oracle Fusion Middleware for Oracle India, to see what his thinking was. I wasn’t dissapointed. As Kumar put it, the idea that SOA requires that participating applications be service-oriented is a “myth.” Most IT shops, in fact, will have a mix of approaches. There will be legacy systems, and there will be “modern” systems, there will be all kinds of middleware and messaging brokers. As he explains it: “If you have an application that is service-enabled, and a whole bunch of applications that are not service-enabled, you can still connect these by deploying adapters. Once [people] realize that, they start to see where SOA can fit in bringing connectivity between diverse transaction engines.” Oracle’s strategy is to position Fusion as the platform that will bring together a lot of diverse assets from across the enterprise into a service layer, and, not surprisingly, this is reflected in Kumar’s statement. But unless an organization throws out all its systems and starts entirely from scratch these days, most SOA efforts will be very ungainly and unique contraptions — and that’s okay. In surveys I have seen and conducted, even the most advanced SOA-savvy companies have less than 20% of their portfolios SOA-ready. And, of course, JBOWS is the predominant architecture at this point. And that’s okay, too. It’s a stage in evolution. And in all likelihood, there will be no compelling need to service-enable 100% of everything. But SOA is in a lot of places, Kumar also reminds us. For example, every time we order from Amazon (an Oracle Fusion customer), the order is processed via a service-oriented framework.>> You can read Joe's blog at: http://blogs. zdnet.com/ service-oriented /?p=1725 Gervas
