Flexible Execution of Distributed Business Processes based on Process Instance Migration

Many advanced business applications, collaborations, and virtual organizations are based on distributed business process management. As, in such scenarios, competition, fluctuation and dynamism increase continuously, the distribution and execution of individual process instances should become as flexible as possible in order to allow for an ad-hoc adaptation to changing conditions at runtime. However, most current approaches address process distribution by a fragmentation of processes already at design time. Such a static configuration can be assigned to different process engines near runtime, but can hardly be changed dynamically because distribution logic is weaved into the business process itself. A more dynamic segmentation of such distribution can be achieved by process runtime migration even without modifying the business logic of the original process model. Therefore, this contribution presents a migration data meta-model for enhancing such existing processes with the ability for runtime migration. The approach permits the inclusion of intensions and privacy requirements of both process modelers and initiators and supports execution strategies for sequential and parallel execution of processes. The contribution concludes with presenting a conceptual evaluation in which runtime migration has been applied to XPDL and WS-BPEL process instances and, based on these results, a qualitative comparison of migration and fragmentation.