Benchmarking a Grid Computing Environment Using Different Communication Bandwidth

Abstract

Internet computing and grid technologies promise to change the way we tackle complex problems. They will enable large-scale aggregation and sharing of computational, data and other resources across institutional boundaries. And harnessing these new technologies effectively will transform scientific disciplines ranging from highenergy physics to the life sciences. In this paper, we construct two heterogeneous PC clusters to form a grid computing environment and setup the middleware - Globus Toolkit on the master node of each cluster. Then, CISCO 2511 router is used to connect these two PC clusters and control the communication bandwidths. Subsequently, we use two different communication styles to benchmark our grid computing system: one is tightly-coupled synchronization job and the other one is looselycoupled communication job. The result show when tightly synchronization application is running with lower communication bandwidth will result in the overall performance down. But the lower communication bandwidth will not affect the performance of loosely-coupled application.