SOA: failure is not an option... yet

IONA's Steve Vinoksi takes me to task for a recent posting, "One Out of Seven SOA Efforts Have Already Failed." The survey was based on a reader poll conducted over at SOA Pipeline.

Steve notes that every SOA-related project he is aware of so far is succeeding just fine, thank you. He writes: "While I don't doubt that there are failing SOA projects out there, the implication of that blog posting -- that it's all somehow caused by SOA itself -- is highly dubious. After all, what percentage of general IT projects fail? Might it be one in seven? Or could it be that the folks whose projects are failing are just blaming it on SOA? How do we know that they even know what SOA is?"

Another stream of reaction to my posting was that the 14 percent reported failure rate is actually very impressive, since the conventional wisdom is that 50 percent of IT projects do not deliver as expected.

Those of you who regularly read this Weblog will probably notice I tend to express both deep skepticism over what is being dished out about SOA, and great hope about the incredible efficiency and new opportunities this new approach can bring us.

Think ahead to ten years from now -- what will conference speakers be saying about SOA? No doubt they'll be reflecting on how overinflated expectations led to many costly failures in the "SOA era." (Web 5.0 and the GagaNet will fix all that, of course.) But the main source of disappointment will likely be that SOA will not have delivered ROI fast enough.

Surveys are imperfect and biased, but they give us a snapshot of what people are thinking and doing. At this point, for the vast majority of companies, SOA is not tangible. In fact, many companies are just getting familiar with Web services standards, and how they can leverage those standards. They are still a step away from synchronizing any Web (and non-Web) services into an SOA.

The bottom line is that it's far too early in the game to call SOA -- or any attempt at SOA -- a failure. But with all the hype-driven expectations, we're setting ourselves up for the inevitable failure pronouncements.

So, we can't be too hasty in pronouncing SOA DOA -- but let's not be too quick to call it the panacea to everything, either. SOA will deliver, but not tomorrow, and not all in one neat package. It will deliver gradually as it moves across business units in the enterprise, in conjunction with other technology and business initiatives. That may be too slow for many people in this fast-moving industry. But we will continue to learn from our failures.