An integration platform for metacomputing applications

Abstract

Simulation and optimisation applications involve a large variety of codes that result in high CPU loads on existing computer systems. Advances in both hardware and software, including massively parallel computers, PC-clusters and parallel programming languages somewhat alleviate current performance penalties. It is the claim of this paper that formal specification techniques, together with distributed integration platforms, provide a sound and efficient support for high performance distributed computing in metacomputing environments. Formal specifications provide rigourous and provable approaches for complex applications definition and configuration, while distributed integration platforms provide standardised deployment and execution environments for coupling heterogeneous codes in problem-solving environments supporting multi-discipline applications. © 2002 Springer-Verlag Berlin Heidelberg.