Theory and Practice

Eric Newcomer (Iona) rightly argues that SOA theory must be grounded in experience. "Real projects, meeting real business requirements every day, are the kinds of things that will make SOAs and ESBs real, and provide the foundations for the real definitions once the hype settles down."

Eric mentions two well-known examples: Credit Suisse and Deutsche Post. A recent book Enterprise SOA by Dirk Slama, Karl Banke and Dirk Krafzig, discusses four detailed examples: Deutsche Post, Winterthur (Insurance), Credit Suisse. and Intelligent Finance.

But I don't think these examples are interesting/complex enough to stretch SOA to its full potential. In order to demonstrate the potential of SOA, you need examples that are reasonably large (not just three boxes joined together) and reasonably complex (with significant transformations and variations, not just "straight-through processing").

Most of the published examples of SOA are small and simple, and invite easy dismissal:

"Why do I need a load of new technology to do this? I could perfectly well build this using my existing software development platform." 

 I am currently working on the hypothesis that the most interesting applications of SOA are going to be found towards the right of the following chart.

Simple Interoperability	Moderate Interoperability	Complex Interoperability
Financial Services Retail SupplyChain Logistics (Basic) Travel	Automotive eGovernment SupplyChain Logistics (Advanced) Telecoms	Aerospace and Defence Consumer Electronics Healthcare and Pharmaceuticals

This is not to say that there aren't any interesting or complex requirements in the financial sector - just that I haven't seen any yet.

One of the implications of increasing complexity is that it places additional challenges on the analysis and design process. (That's why we've been advocating a new approach for business modelling / requirements modelling.)

---

Update (2019)

The CBDI Journal archives are now publicly available:
http://everware-cbdi.com/cbdi-journal-archive
This includes the reports I wrote in 2005, which were advertised in the original version of this blogpost:

• Business Modeling for SOA Worked Example: Pharmaceuticals (May 2005)
• Business Modeling for SOA – Automotive (July-August 2005)

Related posts

Straight-Through Processing (February 2005), SOA Pharma (March 2005), Service-Oriented Business Strategy (May 2005), More Flattery for the Finance Sector (October 2008)


Updated 10 May 2019