Combinational backfilling for parallel job scheduling

FCFS is the most simple, basic and commonly used method of job scheduling in clusters. Backfilling scheduling that small jobs are moved ahead in the schedule can fill the resources gap that is generated by FCFS. However, existing backfilling scheduling algorithms are available for a queued job backfilled to schedule. The resources gap can´t be fully utilized. A method of combinational backfilling for parallel job scheduling in clusters is proposed. It can select multiple jobs combined from the waiting job queue to backfill to maximize the use of idle resources. A comparison with FCFS, EASY backfilling algorithm is given. The results of experiments show that the algorithm proposed can attain the higher utilization of resources in the system than FCFS, EASY backfilling.