A Workflow Platform for Simulation on Grids
Large-scale simulation applications are becoming standard in research laboratories and in the industry. Because they involve a large variety of existing software and terabytes of data, moving around calculations and data files is not a simple avenue. This paper presents the design, implementation and deployment of a simulation platform based on distributed workflows. It supports the smooth integration of existing software, e.g., Matlab, Scilab, Python, OpenFOAM, ParaView and user-defined programs. Additional features include the support for application-level fault-tolerance and exception-handling, i.e., resilience, and the orchestrated execution of distributed codes on remote high-performance clusters.