Darren Wesemann Blog

Yesterday I was in San Jose at the SOA Consortium meetings (nicely organized, I might add, I found the information exchange quite valuable).  At the meeting I was honored to receive an award from CIO Magazine and the SOA Consortium for their 2008 SOA case study, and sat on a panel with my fellow award winners where we discussed our SOA projects and answered some questions from the audience.  Afterwards, in a discussion with an attendee, the topic of data models came up, specifically how they relate to SOA.  The discussion was rather fruitful, so I figured I’d post a summary of the discussion on my blog.

We at SunGard have experimented with data models for years, and with a plethora of applications and components connecting to common data, one could conclude that SunGard might have been taking advantage common structures for years.  To the contrary, only recently have we successfully been able to define and adopt an Enterprise Data Model as an internal standard, but it has been a long-time coming and the road has taught us some valuable lesions, especially as it relates to SOA.

Since SOA is in fact an architecture, and the foundation of any architecture involving software is information, an understanding of the data is going to be important by definition in any SOA.  As it turns out, a semantic structure is actually key to the interoperability of loosely couple components (bound to services and processes) working together in a composite solution.

I get the impression that many lack the basic understanding that both SOA and data models relate to each other in a very dependent way.  I’ve seen some, in fact, “ignore” their data when composing their SOA.

Further, others may leverage a common data model, but a portion of those folks lack the business mapping that a common model requires to be useful in a SOA context.  A useful model needs to mirror the business with readable and clear nomenclature, structure, schema, etc.

After years of collaboration, testing, implementing, etc within SunGard, we’ve achieved this, and are benefitting from it.  Our Enterprise Data Model is implemented in our Stream offering, you can learn more about that here.