The impact of local priority policies on grid scheduling performance and and adaptive policy-based grid scheduling algorithm

Abstract

This thesis addresses a problem with job scheduling in a computational grid. It investigates the performance impact when some sites in the grid apply a priority policy in favor of local jobs and proposes an adaptive site selection algorithm for grid scheduler to reduce the severity of this impact. It is demonstrated that when some sites apply a priority policy in favor of local jobs, other sites will suffer from much longer completion times. The proposed grid scheduling algorithm takes into account local scheduling policies and adjusts the global scheduling accordingly. The results show that the new algorithm can reduce the performance impact due to different local priority policies and perform effectively under various levels of workload and fractions of sites with different policies.