Improving the GridRPC model with data persistence and redistribution

Abstract

The GridRPC model (Nakada et al., 2003) is an emerging standard promoted by the Global Grid Forum (GGF) that defines how to perform remote client-server computation on a distributed architecture. In this model data are sent back to the client at the end of every computation. This implies unnecessary communications when computed data are needed by another server in further computations. Since, communication time is sometimes the dominant cost of remote computation, this cost has to be lowered. Several tools instantiate the GridRPC model such as NetSolve which is a NES environment developed at the University of Tennessee, Knoxville. In this paper, we present the modifications we made to the NetSolve protocol in order to overcome this drawback. We have developed a set of new functions and data structures that allow clients to order servers to keep data in place and to redistribute them directly to another server when needed.