Current thinking about state-of-the-art infrastructure for computational science isdominatedbytwoconcepts: computingclustersandcomputationalgrids.Cl- ter architectures consistently hold the majority of slots on the list of Top 500 supercomputersites, andcomputationalGrids, inbothexperimentalandprod- tiondeployments, havebecome commonin academic, governmentandindustrial research communities around the world. The message passing is the dominant programmingparadigmforhigh-performancescienti?ccomputingonthesearc- tectures. MPI and PVM have emerged as standard programming environments in the...