Experimental analysis of PCPB scheduling algorithm

The problem of energy efficiency appears in every sphere of human activity. At the same time new services require higher processing and transmitting rates. In this paper, the algorithm of tasks scheduling in the server cluster PCPB - Power Consumption and Performance Balance - is presented. It is aimed to reduces the total power consumption of the server cluster without reducing its computing performance. The effectiveness of the algorithm is proved by experiment: an algorithm PCPB is able to increase the performance of the cluster at 3% while reducing power consumption by over 10% in comparison with the widely-used algorithm Round Robin. Existing algorithms of tasks scheduling are analyzed within the paper, including those aimed to increase energy efficiency of the server cluster. One of the key features of the proposed approach is that PCPB involves determining power models for each of the cluster nodes separately. These models are used for the most effective tasks scheduling.