Sunday, 15 March 2009

Is SOA past?

We have spent a lot of effort the last couple of years in introducing service oriented architecture within organization. Some will say that awareness and interest are created, desire and action are dealt with. Best practices will learn how to do a proper SOA project and we can all proceed as usual.

However, there are still a couple of challenges, both from a business and a scientific perspective. Business wise, many organizations may have already implemented SOA internally, but still have to face integration with their business partners. Service Oriented Business is not yet feasible, issues like governance and IT service design for business service support are still challenges.

The scientific challenges are twofold. The one is application wise: how can be model semantics for interoperable business services and can reasoning support mediation functionality between different semantic domains. The second is in runtime service composition and orchestration. Most often, service requirements will not be met one-to-one by service offerings. A runtime mapping needs to be composed, probably based on natural language search technology combined with social tagging mechanisms. These might differ from the ones currently offered by scientific approaches like WSMO. Runtime orchestration means that business persons need to have tooling for composing an orchestration to meet those service composed at runtime.