Dynamic variability support in context-aware workflow-based systems

Abstract

Workflow-based systems are increasingly becomingmore complex and dynamic. Besides the large sets of process variants to be managed, process variants need to be context sensitive in order to accommodate new user requirements and intrinsic complexity. This paradigm shift forces us to defer decisions to run time where process variants must be customized and executed based on a recognized context. However, few efforts have been focused on dynamic variability for process families. This dissertation proposes an approach for variant-rich workflow-based systems that can comprise context data while deferring process configuration to run time. Whereas existing early process variability approaches, like Worklets, VxBPEL, or Provop handle run-time reconfiguration, ours lets us resolve variants at execution time and supports multiple binding required for dynamic environments. Finally, unlike the specialized reconfiguration solutions for some workflow-based systems, our approach allows an automated decision making, enabling different run-time resolution strategies that intermix constraint solving and feature models. We achieve these results through a simple extension to BPMN that adds primitives for process variability constructs. We show that this is enough to eficiently model process variability while preserving separation of concerns. We implemented our approach in the LateVa framework and evaluated it using both synthetic and realworld scenarios. LateVa achieves a reasonable performance over runtime resolution, which means that can facilitate practical adoption in context-aware and variant-rich work ow-based systems.