An intelligent dynamic load balancer for workstation clusters

Abstract
A key issue of dynamic load balancing in a loosely couple distributed system is selecting appropriate jobs to transfer. In this paper, a job selection policy based on on-line predicting behaviors of jobs is proposed. Tracing is used at the beginning of execution of a job to predicate the approximate execution time and resource requirements of the job to make a correct decision about whether transferring the job is worthwhile. A dynamic load balancer using the job selection policy has been implemented. Experimental measurement results show that it is able to improve mean response time of jobs and resource utilization of systems substantially compared with the one without selecting job policy.

This publication has 10 references indexed in Scilit: