Is SOA (Service-oriented architecture) really the magic bullet for large organizations looking to simplify the way they build and integrate information systems or it is just another over-hyped, under-performing acronym designed to keep consultants busy for the next ten years or until the next sure thing comes along?  Accenture, which stands to make billions from the answer to that question, announced yesterday that it will invest $450 million during the next three years to accelerate the growth of its service-oriented architecture (SOA) capabilities

There are lots of definitions of SOA floating around (which is part of the problem) but most people who care agree that it is a modular software design that promises to make business applications easier to write and maintain.  By breaking down a system into a collection of “services” (actually, reuseable, standalone software modules that perform real life business processes, i.e. “credit check”)  SOA promises to provide companies a portfolio of services that can be easily mixed and matched to create automated business processes.  Because the whole system doesn’t have to be re-coded to accomodate new or novel process, proponents claim it can reduce application development time and costs by as much as 50 percent.

One of the early focuses of Accenture’s efforts will be to test the feasibility of building reusable modules by using domain-specific language and model-driven development, in which  applications are written in business language, instead of computer languages. The domain-specific business language automatically converts it into computer code. The language is designed to be useful for a specific set of business tasks.

The new Accenture initiative includes a new R&D lab called the Accenture Technology Lab for SOA Innovation, a blueprint for SOA implementation, and expanding the consulting firm’s  portfolio of cross-industry offerings, especially in healthcare and financial services. The lab will focus on the healthcare industry, with a pioneering project aimed at integrating the multiple steps of filling a prescription, which include credit-checking, billing code generation, and risk assessment. The “e-prescription” project seeks to prove the feasibility of such SOA solutions using software and tools available today.

The press release announcing the project quotes Don Rippert, Accenture’s chief technology officer as saying:  “If this or similar projects aren’t possible, we want our clients to know.”  Rippert’s caveat is something of a disconnect with Accenture’s current marketing efforts which promise, without a hint of uncertainty, that SOA will cut costs, boost ROI and make companies more agile through simplier systems that are easier to build, maintain and integrate.  Put it another way, Accenture is already collecting stud fees for a horse that has yet to leave the gate.

As more and more applications move to the web, SOA is one of those ideas whose logic is unassailable but whose ultimate usefulness remains to be proven.   And it can be extremely expensive.  Just last week the Aberdeen Group released  a benchmark report called Enterprise Service Bus and SOA Middleware which found that migrating to a SOA can eat up 40 per cent of an organization’s IT budget.

You have to give points to Accenture for experimenting with SOA on its own nickel instead of the usual approach of consultants which is to learn on the job while clients foot the bill–although the chances are good that they’re doing that too.