Rickerism

Last week, BPEL4People was formally moved to OASIS. BPEL4People is the right idea, but the wrong execution. There must be a way to integrate people into business processes, but the specification presented to fulfill this goal is far too complicated.

Service-oriented architecture (SOA) provides a breakthrough in enterprise and trans-enterprise integration. SOA has so far been focused on application-to-application integration and, as such, has lacked a certain human element. As efforts in business process orchestration and enterprise service bus expand, implementers are facing the question: what if there is no application on the other side to which to connect?

As unfashionable as it may be to point out, there are still a great number of critical processes in today’s businesses that are not automated. How do we create a complete enterprise service-oriented architecture when many of the services still require human interaction or even human execution? Must we automate every service before achieving results? Is SOA an all-or-nothing proposition like so many failed technology efforts of our past? If so, SOA becomes a burden more than a boon.

What we need is a simple, standard means of integrating into an SOA those business processes that still require human interaction, approval or execution. I call this means Human Services. We must be able to treat Human Services just like other services so that they are interchangeable with other services. The interchangeability not only enables us to combine such manual processes with applications but also allows us to replace the manual processes with an application in the future with no disruption to the business process.

The key failure in Business Process Execution Language and, subsequently, BPEL4People is its lack of simplicity. The approach is overly complicated, offering too many options rather than focusing on a small set of core capabilities. The best way to build complex solutions is to join multiple simple components in various combinations and permutations. Top down definitions by committee are as much a failure in information technology as they are in economics.

I refuse to be just a naysayer. I will not just throw mud on BPEL4People and walk away. I intend to propose an alternative solution on these pages within the month.

References

• OASIS BPEL working group
• OASIS Asynchronous Service Access Protocol working group
• Asynchronous Webservices whitepaper
• IBM BPEL4People papers
• SOA Magazine: Leading Technology Vendors Announce Publication of BPEL4People
• InfoWorld: BPEL4People humanizes SOA