SOA's long, hard slog

The SOA movement, we should recognize, has its share of skeptics. However, the skepticism is more about the pace of SOA change thanthe likelihood of SOA's arrival.

In a post on SandHill.com, Mike Nevins, former managing partner over McKinsey's high tech practice, contends that "widespread deployment" SOA will be slow to come because of "weak business processes" among other problems.

As Nevins puts it: "SOAs will work well when they can automate business processes that: are well characterized; exhibit little or no variability; meet current and growing regulatory requirements for financial control, privacy and the like; and provide appropriate levels of security. Many processes fall short on one or more of these dimensions and are propped up by compensating controls and manual interventions. These band aids work in the circumscribed user communities that use these processes every day. They typically do not scale to the new and less predictable user groups who might access them via SOAs."

Nevins also thinks the conservative resistance of IT organizations (which tend to have pretty staid budgetary and investment practices) and the fragmentation of current data will be problems for the SOA movement.

"That's the bad news," he concludes. "The good news is that all three of these issues need to be addressed anyway. The better news is that tools and approaches to fixing these problems are available and growing in number and use. Process mapping and meta data generation methodologies are being developed and deployed by startups and established players. The leaders in new approaches to running an IT shop have created best practices that others can copy and build on."