Examining the Value of SOA
It does seem about that time that architects weigh in on whether all of the selling and pushing of SOA has indeed been justified by project success.
The verdict, at least from the view of those that have been painting the rosy picture of architectural utopia by the adoption of a brand new encompassing SOA implementation, is less than stellar.
"IT pros have expressed skepticism about SOA’s promised return on investment. A 2007 InformationWeek Web survey of 278 IT pros found that 32% of those using SOA said those projects fell short of expectations. Of those, 58% said their SOA projects introduced more complexity into their IT environments, and 30% said they cost more than expected. Out of all respondents using SOAs, just 10% said the results exceeded expectations [which is confirmed by a survey by] Nucleus Research, which found that only 37% of 106 organizations it surveyed actually were realizing ROI from their investments in SOA technology and programming."
But is the dismal nature of the survey a result of the technology being too young, or being over applied?
To me, at least one factor in SOA projects veering astray is the old elusive yet somehow addictivly overemphasized design principal of "reuse". Huh, I remember when Object Oriented programming was new, and reuse was a goal many were excited about, describing visions of large reusable object libraries. Then there was CORBA, which was going to solve everything, ass well as give us efficiency from reuse. Then J2EE caused many architects to consolidate business objects into a massive singular enterprise layer of reusable services and tables. It was reuse over modularity, ouch. And if we haven't learned our lessons yet, SOA promissed a grand layer of entity services, which could finally, at long last, be what? That's right, reused by multiple systems.
If any design principal is causing waste in the manufacturing process, deemphasize it.
SOA is a great tool. But when it comes to custom business applications, in most cases, SOA is a great tool as a part of a system of cooperative modular efforts to satisfy real business needs.
The litmus test is simple, but often unclear. When talking to an architect, measure the words carefully, and count the sales points geared toward building a great technology vs geared toward engineering a great solution to your business problems.
- mwilson's blog
- Login to post comments