An Energy-aware Task Scheduling Algorithm for a Heterogeneous Data Center

In recent years, rapidly increasing Internet-scale services are deployed in production data centers, which causes a huge amount of energy consumption and environment problem. It is a challenge for these data centers to reduce energy consumption while satisfying the increasing performance requirement of these services. Servers in data centers are usually heterogeneous, which makes task scheduling process more sophisticated. This paper first analyzes and explores two types of heterogeneity from a publicly Google cluster trace. Based on analysis results, we model the task scheduling problem in a heterogeneous data center considering energy consumption and performance requirement, and present an energy-aware task scheduling algorithm based on greedy algorithm. This energy-aware algorithm can achieve local optimal according to our analysis.