I solidly recommend reading the new article ""Business SOA Governance" published by Steve Jones in InfoQ. I believe you will find a lot of information and a vision in this area as well as some fun.
The article starts with the consideration: "If the objective of SOA is to have an IT estate that looks like the business, is managed like the business and evolves like the business then Governance needs to be focused on that goal." Elaborating further in this direction, Steve says: "If the objective is to move towards an IT estate that looks like the business, operates like the business, evolves like the business and is cost in line with the business value it delivers then all aspects need to become Service-Oriented."
This leads to the natural conclusion that the SOA Governance is, first of all, a governance of the transformation program from current IT and Business structural state to the service-oriented state, which I describe in the book ''Ladder to SOE'.
In the line of this transformation, Steve discusses such aspects of SOA Governance as legislative, judicial and operational. It is still a bit unclear to me whether Steve sees the executive aspect to be responsible for the governance enforcement, i.e. governance delivery, or governance is only empowered with the 'sign-off' right for the management enforcement of the governance policies.
There are a few types of SOA Governance exist such as Business SOA Governance, Technology SOA Governance and several sub-types within mentioned areas. In any case, SOA is about Architecture, which means that Business SOA Governance is about delivery of Business Architecture and Technology SOA Governance is about delivery of Technical Architecture aligned with Business Architecture. For instance, Steve says:
"At the transformational level the SOA judiciary, the governance organization, has the role of ensuring that business services and capabilities are delivered correctly. This means ensuring that:
• Business services and capabilities are clearly realized
• Ensuring that people are using the right business services
• Ensuring that rogue services are not developed
• Pollution doesn't happen through tactical decisions
• That funding is shifted to align with the model"
Let me complete my observation with one notification. In both Business and Technical SOA, the borders of the services play a very important role because deliberate of accidental crossing of these borders results in simultaneous violation of several Principles of Service Orientation. Obviously, Steve pays a lot of attention on the role of SOA Governance in policing the service borders. In this context, he states: "What this means is that end-to-end design isn't important, it's about the specification of interfaces at a level which describes...".
In spite of his later explanation of this phrase regarding the judiciary aspect of governance: "The objective of the judiciary therefore is not simply to ensure that a given area delivers its business services to the right specification - it is also there to ensure that one area doesn't care about the internals of another. This is important as it ensures that you don't get implicit dependencies", we know about an amazing ability of technology populists to take things out of the context and make new (but frequently misleading) 'posters' out of them. I would like to worry you from extracting the fragment "What this means is that end-to-end design isn't important, it's about the specification of interfaces" and applying it to an individual service. In the context of governance, mentioned "end-to-end design" relates to the orchestration or aggregation/composition of services, not to implementation of an individual service, IMO.
When one designs a service, there two sub-types of SOA Governance are applicable: governance of the development/implementation and governance of service interaction/execution. SOA Governance of development/implementation may not stop at the level of service interfaces because the essence of the service is not an interaction mechanism but in the service's business functionality and results (Real World Effect) that are provided by the service body (called also service implementation behind the interfaces). Thus, governance of the service implementation is the most important aspect of service realization. However, SOA Governance sub-type related to interaction/execution concerns with only interfaces and SLA. In other words, the end-to-end design of each individual service is not important to the governance of interactions between services, indeed.
I hope you are not confused and will enjoy the article.
Leave a comment