A Resilience Approach to High-Performance Workflows

This report presents an approach to design, implement and deploy resilient distributed workflows. It supports the smooth integration of existing software for simulation applications, e.g. Matlab, Scilab, Python, OpenFOAM, Paraview and application programs. The contribution of the report is a new feature which supports resilience, i.e., application-level fault-tolerance and exception-handling. Connections with exascale computing requirements are also made. An overview of a prototype implementation based on the YAWL workflow management system is given.