My modest proposal for SOA version numbers:
- SOA 1.0 = Spaghetti-Oriented Architecture
- SOA 2.0 = Start Over, All
If an "SOA 2.0" does evolve, it will need to be a dramatic break from the current mode of thinking and deployments.
Richard Veryard may have had this in mind when he became one of the first industry experts to propose the "SOA 2.0" label for the next generation of the architecture. He says its original meaning is lost in current discussions around SOA 2.0.
"The term SOA 2.0 is now starting to be used (particularly by Gartner and Oracle) for a rather confusing technical reframing of SOA. Apparently SOA 2.0 now includes EDA. This surprises many well-informed commentators, who had always thought that SOA was event-driven anyway."
SOA 2.0 -- or whatever comes about next -- will make that definitive break with current-day architectures, one that moves SOA from the internal efficiency stages to actual business transformation, Veryard believes.
As I noted last week on the blogsite, Gartner and Oracle say that the difference between SOA 1.0 and 2.0 is that 1.0 is based on client/server thinking, while 2.0 is based more on event-driven architecture. Gartner's Yefim Natis, for one, is quoted as describing today's SOA as dealing with "a client/server relationship between software modules," with services being subroutines serving clients — not suitable for every business. SOA 2.0 is an event-driven architecture in which software modules are related to business components, with alerts and event notifications. Examples of SOA 2.0 applications would include order processing systems, hospital admissions processes or bank transactions.
Veryard proposed the term SOA 2.0 last October, calling it a way to make the distinction between two manifestations of SOA -- either for internal IT benefits, or for improved business transformation and improved customer experience. (He may have even beaten O'Reilly to this one.)
Veryard's idea of SOA 2.0 breaks out of the box of IT suboptimization, and makes SOA 2.0 real to the business. The Gartner/Oracle vision also looks beyond the IT subop box, but doesn't break 1.0 and 2.0 apart quite as cleanly. You could argue that many SOA 1.0 implementations already are based on event-driven architecture, and are already well beyond anything resembling a "client/server" relationship.
Veryard makes a nice, definitive break between the SOA generations. His proposal is intriguing, as surveys I have conducted and analyzed over the years finds there tends to be two distinct camps around the benefits enterprises expect to see as a result of SOA. One is, through reuse of services and avoidance of more costly enterprise integration projects, the ability to streamline IT development and costs. The other is a higher-level transformative evolution, in which businesses will be able to assemble, disassemble, and reassemble processes to meet new opportunities.
In the case studies and examples I've seen so far, we're very much still sorting through the IT optimization phase of SOA. When large numbers of enterprises start deploying SOA to move the business itself forward, perhaps that will be a sign we're entering the next stage -- whatever it will be called.