A PROVABLE EFFICIENT CONTROL ALGORITHM TO REDUCE DATA CENTERS POWER COST WITH ON-SITE RENEWABLE ENERGY GENERATION

摘要

Cloud computing and big data era introduces a new opportunity for enterprises to commission their massive computing tasks to Internet data centers (IDC) with relatively low cost. However, the explosive growth of massive data processing has imposed a heavy burden on computation, storage, and communication in IDCs, which hence incurs considerable operational expenditure to data center providers. Data in the cloud computing need to be transferred and processed with high efficiency and cost effectiveness while meeting the demand on reliability. This paper formulates the cost minimization of IDCs servers and renewable power problem to minimize time-averaged expected power cost subject to the constraints of data queue stability and server allocation policy with the consideration of renewable energy utilization. We design an algorithm leveraging the Lyapunov optimization technique to approximately solve the CMISRP problem. All data are originally from Ordos Uni-Cloud Technology Co., Ltd. to simulate the proposed algorithm. Numerical results show that our algorithm can reduce total energy cost while guaranteeing QoS, queue stability and server allocation policy.