A couple of things about IBM’s October SOA announcement caught my eye.
The first was the introduction of IBM’s Registry/Repository, WebSphere Service Registry and Repository. At last IBM has filled this gap in its portfolio, but the point that interests me is that IBM has chosen to go the combined route, rather than the spearate offerings one used by BEA for example. For those not into the nuances of the wording, Repository is supposed to be the design-time store and Registry refers to the run-time store. For ages I have thought of all this as ‘The SOA Registry’….. but I have been educated!
Or have I? My own opinion is it is silly to have these two components separate – it adds an extra layer of complexity, always a bad thing in my mind. The counter-argument is apparently based around the fact that the needs of the two environments, design and runtime, are different (eg performance) and therefore putting them together is a problem. This sounds lazy to me – I am sure these things can be sorted by the software. I think this is small beer compared to the confusion generated from having two different components.
The other point I found interesting was focus around offering industry-verticalised SOA tools, with theidea being presumably to enable industry vertical business analysts to easily create their own business services. A fascinating idea, but whether IBM’s WebSPhere Business Services Fabric (based on teh Webify acquisition) can deliver it depends largely, I suspect, on getting third parties to play. So far, there appears to be some solutions for Healthcare and Insurance, but unless the industry vertical application vendors decide to play, bringing their extensive domain expertise, the cupboard may remain pretty bare.