Improving Processor Allocation in Heterogeneous Computing Grid through Considering Both Speed Heterogeneity and Resource Fragmentation

In a heterogeneous grid environment, there are two major factors which would severely affect overall system performance: speed heterogeneity and resource fragmentation. Moreover, the relative effect of these two factors changes with different workload and resource conditions. Processor allocation methods have to deal with this issue. However, most existing allocation methods focus on one of these two factors. This paper first analyzes the relative strength of different existing methods. Based on the analysis, we propose an intelligent processor allocation method which considers both the speed heterogeneity and resource fragmentation effects. Extensive simulation studies have been conducted to show that the proposed method can effectively deliver better performance under most resource and workload conditions.