When is it safe to say SOA is successful?

According to a recent survey by SOA governance company Amberpoint, we can say so now. The survey found that nearly four out of 10 enterprises report their SOA implementations met all their goals. Sixty percent said SOA met most of their goals.

But even more telling is this statistic: Only 1.5 percent of SOA projects ended in failure, a figure that beats the industry average for all software projects, according to TechWorld.

Even if you add in those surveyed who opted not to proceed with their SOAs, the failure rate still isn't even 5 percent.

This surprised Amberpoint as much as the rest of us. "Ed Horst, Amberpoint's VP of marketing, was surprised by this particular result and could offer no explanation for it," the article notes.

Apparently, the SOA governance company wasn't anticipating these type of results, since the survey didn't push for more details on the SOA implementations. Still, there's enough here to answer a lot of "Yeah, but" arguments you might have. For instance:

• The survey did reveal that only 20 percent of those surveyed were restricting the SOA projects to a single department -- so small implementations are not a huge factor in the success rate.
• The survey also showed 58 percent of enterprises included "non-SOAP" messaging, such as MQ or RMI, in their SOAs, indicating this isn't a case of "Just a Bunch of Web Services" masquerading as a SOA.
• The survey did show that packaged applications -- specifically, the article mentions SAP - are included in 68 percent of SOA systems, so while you may not be able to "buy" a SOA, you can certainly buy a solution that helps move your SOA along.

While the Amberpoint survey suggests SOA is a success, others are less keen to embrace SOA's success.

For instance, Oracle's Larry Ellison said it could be 10 to 20 years before companies can take full advantage of SOA's benefits -- although, he made those comments before Oracle's big move into SOA-based data integration suite this month.

Australian Oracle consultant Chris Muir is inclined to agree with Elison. As Muir wrote recently on his blog, in the real world, building a SOA can be very messy. Although his problems actually have more to do with wayward business processes than anything specifically SOA, his point should be well-taken: SOA may be a good approach, but it's not magic. SOA still has to work in a world that may not be ready to capitalize on what it offers.

Wall Street also remains skeptical, according to this blog post from Wall Street & Technology.

Wall Street is bothered by the lack of service level agreement and the resulting service quality assurances, according to Hugh Grant, director of global IT research and development at Credit Suiss.

That's hardly not surprising from a conservative and heavily-regulated industry that measures success by minute market fluctuations. And who can blame them? Grant also listed concerns about security, funding, risk and compliance as major obstacles to SOA adoption at investment firms.

These companies are exploring SOA slowly, though their use is restricted primarily to small integration and business activity monitoring projects within business units, according to Grant.