Thinking about Gartner's well-known technology hype cycle, I'd argue we're approaching the "Peak of Inflated Expectations" when it comes to service-oriented architecture (SOA) and business process management (BPM). Adoption is still on the upswing, and plenty of companies have realized moderate wins or even big gains through SOA integration, streamlined processes or both. What will inevitably follow is the "trough of disillusionment" in which critics will stir an SOA and BPM backlash. One of their chief complaints is likely to be the chaos that will ensue as service and process components multiply and companies start to loose track of how they all fit together.
SOA and BPM should be discussed together because they are "joined at the hip," as Derek Miers argues in an upcoming white paper. "Common to both approaches is the concept of loose coupling, which enables both internal and external applications to be spread across a distributed technology infrastructure," Miers writes."They both encourage reuse... [and] can adapt in line with changing needs of the prevailing competitive environment, rather than being constrained by brittle, tightly integrated applications. The response to a given business need is composed (from loosely coupled components) rather than tightly integrated at design time. If approached correctly, the cost of development and maintenance is considerably reduced."
What has yet to be seen among many SOA — and BPM-adopting organizations is whether they can make the transition from "once-and-done IT project" thinking to an ongoing, holistic and enterprisewide approach to these disciplines.
"Both BPM and SOA could be thought of as a way of thinking about how the business and governance model should be designed and a way of delivering the technology and applications to support that design," Miers says. "Both concepts involve a journey, not a destination. At their core, both employ an iterative approach to business performance improvement."
It's the iterative part that many companies just aren't practicing. They're not "closing the loop" by using their experience (or, better still, metrics) to tweak services and processes for optimal performance. Rather, too many are redesigning once and declaring victory because they've realized a few gains. Needs change, however, and the trough may come when components start falling out of tolerance because they're not periodically revised.
If companies are practicing continuous improvement, and they're also taking their SOA and BPM initiatives enterprisewide, the disillusionment may arise as service and process components multiply — both in total number and in versions.
"The very feature that makes BPM so attractive to businesses — the ability to chop and change processes at the drop of a hat — makes it dangerous in the medium term and possibly fatal in the long term," argues BPM expert Keith Harrison-Broninski. "We are heading for a process meltdown as processes spawn one another without limit and without any overarching means of control."
Many vendors and practitioners do recognize the potential for service and process chaos, and my experience with content management technology leads me to believe that it's not an insurmountable problem. If repositories can manage the complexities of, say, an FDA drug submission or a Boeing aircraft manual, then registries and SOA/BPM repositories can handle proliferating services and process components. As explained in "Analysis: When SOA and Process Management Merge", Software AG and Fujitsu are tackling the problem with their jointly developed CentraSite "registory," which combines the features of a registry and repository. Systinet and others are addressing this looming problem as well.
The lesson of the Hype Curve is that technology is never as easy as the early proponents would have you believe. It's time to prepare for the long haul and to think of and practice SOA and BPM as ongoing approaches rather than "that's done" projects.