Psst ... got a SOA Reference Model? Want another one?
A recent podcast on SOA has resulted in some interesting debate within one OASIS technical committee. In his podcast David Linthicum discusses the Open Group's efforts to define a SOA reference model yet does not mention the work conducted over the past 2 years by the OASIS SOA-Reference Model (SOA-RM) technical committee. This is a shame because the OASIS group have done really well in generating consensus around an important and often fractioned subject. However, as Jeffrey Estefan points out, it's not just the oversight of an entire standard's committee's work that people should be aware of, there are fundamental differences of opinion:
One troublesome item of note is that David blended the notions of "reference model" with "reference architecture," which we all know are not synonymous. And the podcast diverged into a discussion comparing and contrasting SOA with EA (Enterprise Architecture), probably based on the fact that debate on The Open Group SOA RA work was taken in the context of EA. Unfortunately, too many people don't know that EA really is and that fact that EA is not an architectural style or architectural paradigm like SOA is but rather that EA is an essential element of IT governance and is used as an strategic planning tool used to help make IT investment decisions. And that it is comprised of a current ("as-is") state, a desired target state ("to-be"), together with a transition plan to help define how to migrate from the current state to the target state. And that it should be updated on annually.
However, maybe the most important question that arises from all of this is: just how many reference models do we need for SOA? Although not identical, there was an attempt by the W3C and member organizations back in 2002 to define a Web Services architecture, but that foundered due to a number of reasons, one of which was the annual "debate" on REST versus Web Services. Then the OASIS SOA-RM committee began, though unfortunately without the help of vendors such as IBM and Microsoft. Is this latest effort yet another example of the vendor wars that have wracked Web Services for the past 5 years and resulted in FUD (Fraction, Uncertainty and Doubt)? Or perhaps there are fundamental differences between the Open Group and OASIS efforts? Given the depth and detail of the SOA-RM work, this latter explanation does seem difficult to justify. Rather than re-inventing the wheel, maybe the Open Group should reach out to the OASIS SOA-RM technical committee and feed into their effort?