Run-time Optimisation of Grid Workflow Applications

The execution of workflow applications on the grid is a complex issue because of its dynamic and heterogeneous nature. While the grid provides good potential for achieving high performance, it also introduces a broad set of unpredictable overheads and possible failures. In this paper we present new methods for scalable and fault tolerant coordination of workflows in dynamic grid environments, including partitioning, static and dynamic optimisation, as well as virtual single execution environment, incorporated into the ASKALON distributed workflow enactment engine. We demonstrate the effectiveness of our methods on a material science workflow application executed in a real-world grid environment