Tony Baer has an interesting post at http://www.onstrategies.com/blog/?p=251, ‘SOA in a Recession?’. The question here is ‘what will SOA investments look like during a recession?’ Having been a part of the big client server moves of the mid to late 1990s, web deployments of 1997 to the present, and someone who has done training on SOA across the country for Wintellect, I have to say that SOA feels different from the previous two items. For client server, we had to start thinking about our applications differently. Bits of the application lived in different processes on different machines. Here, we had to rearchitect applications to deal with a new security model and to deal with the greater latencies involved in method and database calls.
The web-based applications extend those latencies all the way out to the user interfaces and brought about new design issues. Web designers have to worry about browser compatibility, screen refreshes, etc. This change was easier than client server because we already knew how to handle latencies. The web only required rewrites of the user interfaces. Back-end tiers could be left alone.
SOA is even gentler. The SOA experiments in flight at this time all revolve around exposing existing client server deployments to new platforms. Select subsystems are being exposed to new development via Web services. The discussion folks are having is more like ‘let’s put a Web service wrapper around system X’ instead of ‘let’s rewrite for SOA.’ I see SOA as being an enabler for continued development even in the face of a recession since it allows one to reuse existing code investments with minimal effort. On my current project, we are regularly exposing data between the Java and .NET worlds via Web services. Coding, testing, and deploying the Web services is usually a sub 1 hour task.