Coordination for fragmented loops and scopes in a distributed business process

Abstract

As partner relationships become more dynamic and global boundaries give way to a more agile and dynamic environment, the ability to distribute one's processes in an agile manner becomes increasingly important. Such processes may need to be split not only along their explicit dependencies but also along more complex behavior such as recovery behavior and loops. The resulting process fragments can be distributed and wired together, recreating the execution semantics of the original process model. In earlier work, we presented BPEL fragmentation covering data and explicit control dependencies. We now extend the approach to handle fragmenting loops and scopes. Maintaining the focus on standards and maximizing extensibility of Web service runtimes and standards, the solution defines and uses two new coordination protocols that plug into the WS-Coordination framework. The approach uses the standards as much as is feasible and addresses the remaining required functionality by providing architected extensions. This results in layered approach that maximizes transparency and interoperability. After defining the fragmentation approach for scopes and loops, an implementation is presented that extends the Active Endpoints BPEL engine and a WS-Coordination system. A detailed example is used to illustrate how the protocols are used at runtime to enable the coordinator and the process fragments to recreate the behavior of the original, unsplit process model.