Unlike the November elections, the returns aren't in for service-oriented architecture (SOA). According to a survey conducted by Nucleus Research [2] and KnowledgeStorm [3], only 37 percent of enterprises that adopt SOA reap a positive return on investment. Nearly two out of three see no net payback; however, the survey also revealed that the problem isn't technical and instead involves how SOA is used. That's good news because some System i users and applications may be positioned to benefit from a well-implemented SOA strategy.
There's a basic reason SOA lacks ROI, says David O'Connell, a senior analyst with Wellesley, Massachusetts–based, Nucleus Research. "It gets adopted and remains isolated on a project-by-project or team-by-team or department-by-department basis."
The result is that there's not much breadth of reuse of code by developers. According to the survey, there is an increase in developer productivity by an average of 28 percent with SOA; however, without widespread code reuse, this productivity boost alone doesn't put SOA enough in the black to make its widespread deployment worthwhile.
As for why SOA is underutilized, O'Connell notes that the availability of services has to be broadcast, the use and reuse of services controlled, and the repository for services known. All of these factors point to a centralized SOA effort, perhaps along with a culture that pushes its use. If SOA is initiated at the department, team, or project level, those same factors may tend to keep the technology at that level.
For System i users, this situation means that SOA may or may not make economic sense. The difference between the two outcomes will largely be driven by conditions within the enterprise and not the technology. On one hand, the small and medium business (SMB) market that many System i boxes inhabit isn't conducive to making SOA pay off. Brian Jeffery, managing director of the Los Angeles–based International Technology Group [4], notes that vendors often implement service-oriented architecture in their products. Thus SOA is used within organizations and provides a return on investment almost invisibly.
However, those in the SMB arena doing custom development may have only a few programmers, with the total perhaps in single digits or at most a dozen. These resources are often busy with something other than the development of a service that may perhaps be reused at some later date. "You're typically working on much more sort of practical, short-term kinds of things," says Jeffery.
Another barrier, he points out, involves the level of sophistication of the developers. Implementing SOA can be a challenging process, and doing so successfully requires expertise in the right areas. Given the lack of programmer availability and the need for somewhat sophisticated technical knowledge, it may not make sense or even be practical for smaller organizations to deploy service-oriented architecture.
Charles King, principal analyst of Pund-IT [5] of Hayward, California, notes that the SOA success stories he's aware of involve large organizations where the culture needed to make SOA pay off can thrive -- and the resources to implement the approach are present. That combination likely explains those successes.
King likens SOA to the compound interest banks and other financial institutions pay their depositors. The accrual of interest, in this analogy, is the code that can be reused. The payoff from SOA, like interest, doesn't come quickly and may take some time, another factor that could prevent success in SMBs.
There may be, however, some areas in which the System i is well positioned to take advantage of SOA and where the use of SOA in small and medium-sized businesses might become common. For example, the Nucleus Research survey found that healthcare organizations were among SOA power users, with 62 percent of developers reporting adoption.
King notes that the System i has deep roots in healthcare. What's more, he adds, trends in the field seem tailor made for SOA. There's a move toward a more systemic approach to application development and toward multiple applications leveraging common databases. What's more, the major players in healthcare are all trying to gain every possible benefit of scale.
"That type of aggressive business market might spark the kind of benefits you'd see from SOA, and I would think that the System i would have a play there," Kings notes.
Links:
[1] http://systeminetwork.com/author/hank-hogan
[2] http://www.nucleusresearch.com/
[3] http://www.knowledgestorm.com/
[4] http://www.intertechnogroup.com/
[5] http://www.pund-it.com/