Chapter Seven. The Workflow Management Coalition (WfMC)

ALTHOUGH ITS INFLUENCE HAS WANED IN RECENT YEARS, the Workflow Management Coalition (WfMC, http://www.wfmc.org) remains a rich source of ideas about BPM. The contributions of this group include a decade-old workflow reference model and specifications for the design of its main pieces, including a client API and an exportable process definition format.

The WfMC, founded in 1993, is a nonprofit organization that writes workflow standards, including a workflow reference model and specifications for each of its core interfaces: XPDL, WAPI, and WfXML. The WfMC also produces voluminous BPM literature; notably, the annual Workflow Handbook (http://www.wfmc.org/information/info.html). The group is chaired by Jon Pyke and has more than 300 member organizations, including BEA, FileNet, Fujitsu, Hitachi, IBM, NEC Soft, Oracle, Sun, TIBCO (Staffware), Toshiba, Vignette, Vitria, and WebMethods.

This chapter highlights the core WfMC design ideas. The first section examines the WfMC reference model. Next, it explores XPDL, providing an extended example, a brief look at the main elements, and a summary of its support for the P4 patterns. The next section examines the major functional elements of the WAPI and provides an example. The chapter concludes with a brief look at the WfXML.

WFMC STILL MATTERS

In the fast-changing world of BPM, WfMC is an older, more established group with mature ideas. With the increasing popularity of the competing BPEL language (Chapter 5), the WfMC must somehow justify its relevance. The trick is to not compete with BPEL but complement it, as the BPMI has done (Chapter 6) by surrounding BPEL in its proposed BPM stack with BPMN (modeling notation) and soon-to-come specifications BPQL (query language for process monitoring), BPSM (common process metamodel), and BPXL (BPEL extensions). Significantly, the BPMI excludes its own execution language BPML from the stack, accepting BPEL's reign in that category. BPMI seems stronger and fresher than ever because it is building key pieces of a platform that solution providers will actually use.

WfMC's predicament is similar. Despite having a good reference stack with substantial implementations of each the major interfaces, one of those interfacesnamely, XPDLis a competitor to BPEL. To be successful, the WfMC should drop XPDL from the stack in favor of BPEL and then demonstrate the value of the overall model and its non-XPDL interfaces. That model, whether it includes XPDL or not, is full of good, current ideas. But leaving XPDL in makes the WfMC seem as if it is working in a vacumn, with an obscure process language that no one else is using.