Optimizations of data distribution localities in cluster grid environments

Abstract

The advent of widely interconnected computing resources introduces the technologies of ubiquitous computing, peer to peer computing and grid computing. In this paper, we present an efficient data distribution scheme for optimizing data localities of SPMD data parallel programs on cluster grid, a typical computational grid environment consists of several clusters located in multiple campuses that distributed globally over the Internet. Because of the Internet infrastructure of cluster grid, the communication overhead becomes as key factor to the performance of parallel applications. Effectiveness of the proposed distribution mechanism is to reduce inter-cluster communication overheads and to speed the execution of data parallel programs in the underlying distributed cluster grid. The theoretical analysis and experimental results show improvement of communication costs and scalable of the proposed techniques on different hierarchical cluster grids. © Springer-Verlag Berlin Heidelberg 2005.