面向需求响应的数据中心供能系统优化调度

Optimal dispatching of data center energy-supplying system with demand response

数据中心作为高能耗和高碳排放量的互联网基础设施,国内外数据中心运营商通常采用分布式可再生能源电站为其供电。基于数据中心工作负载的可调度特性,提出数据中心电/冷负荷的协同管理方式及模型,在含分布式光伏发电的数据中心供能系统下,以日购电费用最小为目标,兼顾服务器集群运行可靠性需求,建立分时电价环境下面向需求响应的数据中心供能系统优化调度模型,并采用分支定界法求解。基于中石油数据中心一期工程数据和谷歌公布的工作负载样本数据集的算例仿真验证了所建模型的可行性和有效性,结果表明该模型最高可削减11.5%的数据中心购电费用,提高了可再生能源利用效率。

Distributed renewable energy power station is usually utilized to supply power for the data centers by reason of data centers＇ characteristics of high energy consumption and high carbon emissions.In this paper,a cooperative management scheme and a collaborative management model with an electric/cooling load of data center are proposed by optimizing the execution time of workloads.Based on the data center energy-supply system containing the distributed photovoltaic power generation,an optimal dispatching model of data center energy-supplying system with electricity demand response is established under time-of-use electricity price environment.The objective function of model is the minimum of the daily electricity bill of data center.The restrictions of the model include operational reliability,requirements of the cluster server and so on.Besides,the branch and bound method is employed to solve the model.The simulation based on the data of first phase in CNPC data center and Google cluster-usage traces shows the feasibility and effectiveness of model.The results show that the utilization efficiency of renewable energy is raised and the electric bill is reduced up to 11.5%.