Cost Reduction for Micro-Grid Powered Data Center Networks with Energy Storage Devices

The data center networks including multiple geo-distributed data centers begin to surge for meeting the ever-increasing Internet demand. To reduce the electricity bills in data centers, the current efforts center on developing sustainable data centers and improving energy efficiency. In this paper, we power data center with a microgrid which have conventional generators (CG), renewable energy sources (RES) and connected with electricity markets. To smooth the RES uncertainty, we integrate energy storage devices into microgrid. Besides, we consider any kind of energy can be charged into batteries or directly cover the demand of data centers. A stochastic program is formulated by integrating the geo-distributed load balancing, the energy management in microgrid and server configuration while guaranteeing the quality of service experience by end users. We design an online algorithm using Lyapunov optimization techniques without foreseeing any future information. The numerical evaluations based on real-world data sets corroborate the superior performance in reducing cost compared with previous works.