Friday, November 21, 2008

SOA slowing down

Gartner says the "Number of Organizations Planning to Adopt SOA for the First Time Is Falling Dramatically". I will summarize the facts as below:

  • 53% of person contacted are using SOA in some part of their organizations
  • 25 % were not using it but had plans to do so in the next 12 months
  • 16 % no plan to use it
  • 20 % are building Event Driven Architecture
  • 20 % are planning to do EDA in the next 12 months
  • Since the beginning of 2008, there has been a dramatic fall in the number of organizations that are planning to adopt SOA for the first time. down to 25 percent from 53 percent in 2007
  • Many organizations have evaluated SOA and have chosen not to spend time and effort on it
  • The highest concentrations of organizations not pursuing SOA and having no plans to do so are in process manufacturing and agriculture and mining
  • SOA adoption in Europe is nearly universal, moderate in North America and lagging in Asia
Main reasons listed are:
  • No clear business case – but there is a great deal of confusion about how to construct a business case for SOA
  • Lack of skill and no plan to acquire necessary skill
  • Use of modern programming environments is closely associated with SOA
  • Legacy skill + SOA are scared

In my recent work on the book "Resilient Information System" as co-author, I have to study what are the pains and use cases our BRMS customers have in term of SOA. I'm still doing interviews on that matter but basically there are four main subjects to address when migrating to SOA :
  • Data: work on the definition of domain data model, or put in place some transformation layer to present the different view of the data to the applications
  • Business rule: be able to understand, to externalize and to easily change the business rule outside of the applications.
  • Expose functions as business services: once reusable points have be identified, the services can be designed and documented (WSDL) for reuse.
  • Re-engineer the business processes to support automation using reusable services.

The successful deployments among our customers are done in that order. If CIO buys upfront the full SOA suite, he will most likely not get the bang for the money, as the migration to an Extended SOA is long. So the business case has to be built by increment with the first goal of deploying tools that empower the business user to change his business policies. When you are successful on this matter, the other business cases are easier to articulate, because business users are seeing the value. As explained in the Agility Chain Management System the core technologies to support the implementation are MDM, BRMS and BPM. Deploying SOA without BRMS and MDM does bring to the red path of the SOA maturity matrix. BPM, as BPEL orchestration engine, is not the solution to buy upfront, it will arrive later in the SOA maturity when we have reusable services that we can orchestrate.

SOA is a viable architecture design and approach as was OOD in programming, so please do not kill it. Now it has to be decided as a corporate initiative and implemented by increment, project after project.