Computer cluster scheduling algorithm based on time bounded dynamic programming

Computer clusters are currently the most used distributed computer architecture. Efficient utilization of computer cluster depends on a scheduling policy that is applied. Scheduling of jobs in computer cluster is a complicated task due to frequent changes in the workload. In this paper we present scheduling algorithm that is based on EASY backfilling scheduling strategy. Dynamic programming with time restriction is used to calculate as good schedule as possible within given time constraints. Algorithm is evaluated on several computer cluster workloads and is shown to outperform original backfilling strategy.