Saturday, June 13, 2009

SOA Definition Revisited

It is time to revisit the definition of SOA. In an earlier posting I provided the following definition and suggested it may evolve.

SOA is a software architecture where the business logic is encapsulated in services and a single application may be distributed over many processors.

A service in this context is a software component which·

  • Hides its implementation,
  • Is based on standards,
  • Is location transparent,
  • Is message coupled, and
  • Is accessible through defined platform independent
    interfaces.
I recently came across this definition in a book published by IBM Press (see reference below).

A service-oriented architecture is a framework for integrating business processes and supporting IT infrastructure as secure, standardized components—services—that can be reused and combined to address changing business priorities.
Beiberstein et al refer to SOA as a framework I refer to it as an architecture which should be self evident. I do not think this is an important difference. The difference suggests another layer of abstraction a littler further away from the implemented product. The Zachman framework is even more abstract though.

There is an IEEE definition for architecture it is "the fundamental organization of a [business] system, embodied in its components, their relationships to each other and the environment, and the principles governing its design and evolution". I think this sufficiently describes what is needed for an SOA without out going to the next level of calling it framework. I will stick with calling SOA an architecture for now, although I acknowledge that there are a number of technical, logical, and business centric models that should be documented for an SOA.

I like the reference to business processes and I compare it to my discussion of business logic although business logic is at finer grain and is within a service. There is an interesting argument that Business Process Modelling (BPM) should be separated from SOA but generally a good SOA implementation will allow an organization to implement its business process more easily than with other architectures. The BPM tool may be considered to be outside of the services but still be part of the overall SOA. I'll watch this space and consider where I draw my SOA boundary.

Standardize components (services) are obviously central to SOA. I don't think they are necessary secure components. Unsecure services can still be SOA even though it may be bad SOA and security has been a bit of a challenge for the evolution of SOA.

Reuse and combination are important outcomes of SOA but they do not necessarily define SOA.

My SOA definition above is still looking strong. It has the possible deficiency that it uses the term "Message Coupled" rather than "Loosely Coupled". "Message Coupled" is not in wide usage. While "Loosely Coupled" is in wide usage and is descriptive to a point, I found it too loose in my earlier posting and still stand by it.

References
Norbert Bieberstein; Robert G. Laird; Dr. Keith Jones; Tilak Mitra, Executing SOA: A Practical Guide for the Service-Oriented Architect, IBM Press: The developerWorks® Series, May 2008, Page 4.

The definition of Architecture can be found in IEEE Std 1471-2000 Systems and Software Engineering—Recommended Practice for Architectural Description of Software-intensive Systems [2007-07-15]

1 comment:

chinesische medizin said...

A service-oriented architecture is the underlying structure supporting communications between services. SOA defines how two computing entities, such as programs, interact in such a way as to enable one entity to perform a unit of work on behalf of another entity. Service interactions are defined using a description language. Each interaction is self-contained and loosely coupled, so that each interaction is independent of any other interaction.