Workflow management systems (WfMSs) frequently use data to coordinate the execution of workflow instances. A WfMS evaluates conditions defined on data to make the control flow decisions i.e. selecting the next activity or deciding on an actor. However, data - within and outside of a running workflow instance - may change dynamically. Modifications of data needed for past control flow decisions may invalidate these decisions. We analyze the desired synchronization policies, and propose a mechanism called data guard to selectively guarantee that significant changes in data are recognized and handled by the workflow management system to ensure correctness of workflow execution in face of asynchronous updates.

Top