A Combined Approach for Implementation of Broadcast Operation in Grid Computing

Abstract

In distributed computing, the collective communications scheduling is among the most important scheduling problems related to intensive applications executed over heterogeneous platforms. The optimization of collective operations allows the improvement of parallel and distributed applications performance by reducing the completion time of these operations. In this paper, the authors are particularly interested by the optimization of broadcast operation executed over large scale distributed environment such as grid computing. For this aim, we combined the two levels approach implemented in MagPIe library proposed for the hierarchical large scale systems with the ECEF (Earliest Completion Edge First) heuristic proposed for the IPG (Information Power Grid). Simulation results show the advantage of our proposed hybrid strategy compared to the classical ECEF heuristic.