I noticed that my blog has started off very Process Server and Integration developer centric. I’m sure this is because I’m currently working on these products at my client (along with being a member of the development team). I just wanted to point out that I’m also conversant in WebSphere Business Services Fabric (WBSF); 6.0. I had the opportunity about 8 months ago to be the ‘first’ non-IBMer to actually run a scenario end-to-end using that product stack Webify before IBM bought them). I like to think I was the first because I did it back when the IBM.com site itself didn’t even mention the product existing 🙂
At a high level overview, WebSphere Business Services Fabric lets you dynamically route web service requests between multiple endpoints. The canonical example being routing a request from a ‘Gold’ member to a premium service while ‘Silver’ members go to a lower quality one. Fabric is built on top of Process Server and Integration Developer, creating a complete integration environment that can deal with the complete SOA life-cycle. The neat part about fabric is that it allows you complete dynamicity on the fly. You can change the routing rules in an admin console-like front-end while maintaining a business oriented view of the data as opposed to a low level XML or XSD one. It also integrates with WebSphere Services Registry and Repository which will allow you to track service usage, versioning etc.
Fabric uses the concept of Ontologies to define the business data flowing through the system. IBM provides some content packs for the Insurance and Heath care industries that are great for getting to know the product. Unfortunately, back in 6.0, an ontology editor wasn’t included so you had to go with an open source solution which was difficult.
When I was dealing with the 6.0 release, things were pretty difficult to understand given the new concepts and programming style of the product. I see now that WBSF v6.1 has been released. I’m sure that after a good 8 months of development times, things have gotten better.
My only issue with Fabric is that it’s a bit of a product before it’s time. Its used to dynamically select an endpoint when multiple endpoints that perform similar units of works exist in an enterprise. In my experience, the SOA Industry isn’t quite at that stage. What I see is mostly “Lets create some services because we don’t have any” rather than “Oh my.. How do I maintain control over the ten thousand services in my enterprise”. But a few years down the road, when services are being consumed by services that are being consumed by more services and a low level service need to be introduced, Fabric will be a great solution to choose when to call version 1.1, 1.2 or 2.0.