Load Balancing for Minimizing Execution Time of a Target Job on a Network of Heterogeneous Workstations

A network of workstations (NOWs) may be employed for high performance computing where execution time of a target job is to be minimized. Job arrival rate and size are “random” on a NOWs. In such an environment, partitioning (load balancing) a target job based on only the first order moments (means) of system parameters is not optimal. In this paper, it is proposed to consider the second order moments (standard deviations) also in load balancing in order to minimize execution time of a target job on a set of workstations where the round-robin job scheduling policy is adopted. It has been verified through computer simulation that the proposed static and dynamic load balancing schemes can significantly reduce execution time of a target job in a NOWs environment, compared to cases where only the means of the parameters are used.