WCSE

GBESS: An Energy-Saving Cluster Scheduling based on Grouping and Balancing

Abstract— With the rapid development of Big Data and IoT, the scale of server cluster in enterprise data center is increasing sharply. In order to deal with the issue of energy consumption, this paper proposes a Grouping and Balancing based Energy-Saving Scheduling (GBESS). GBESS is a double level strategy: 1) It reduces energy consumption by grouping the nodes of cluster into hot group and cold group; 2) It creates slack space for energy-saving grouping and ensures the lower bound of QoS by load balancing the hot group. Experimental evaluation shows that GBESS can reduce the energy consumption by 12.1% on average without sacrificing QoS. Furthermore, the energy efficiency of GBESS can improve along with the cluster scale growing.

Index Terms— Energy-saving, Scheduling, Load balance

Shang Qiuli, Liu Bingshuang, Tu Bo, Dai Shuaifu, Zhang Jianyu
Seven North Road No.42 Building, CHINA

[Download]

Cite: Shang Qiuli, Liu Bingshuang, Tu Bo, Dai Shuaifu, Zhang Jianyu, "GBESS: An Energy-Saving Cluster Scheduling based on Grouping and Balancing," Proceedings of 2017 the 7th International Workshop on Computer Science and Engineering, pp. 789-793, Beijing, 25-27 June, 2017.