Investigation to make best use of LSF with high efficiency

A network of commodity workstations offers an effective platform for large scale parallel processing. The effectiveness of such systems can be greatly improved by efficient cluster management software. In this paper, we focus upon the widely used load sharing facility (LSF), and investigate its ability to run both parallel and serial simulations and increase the throughput of jobs. The experiment shows that LSF does a good job in terms of balancing the load for serial jobs and avoiding the machines with high processor utilization rates. Using the result of the experiment, we created a useful diagnostic tool for assessing the impact of load balancing software.

[1]  Miron Livny,et al.  Condor-a hunter of idle workstations , 1988, [1988] Proceedings. The 8th International Conference on Distributed.

[2]  David A. Lifka,et al.  The ANL/IBM SP Scheduling System , 1995, JSSPP.

[3]  M. A. Pettipher,et al.  The Development of an MPP Implementation of a Suite of Finite Element Codes , 1997, HPCN Europe.