Online capacity provisioning for carbon-neutral data center with demand-responsive electricity prices

Abstract

Due to the huge electricity consumption and carbon emissions, data center operators have been increasingly pressured to reduce their net carbon footprints to zero, i.e., carbon neutrality. In this paper, we propose an efficient online algorithm, called CNDC (optimization for Carbon-Neutral Data Center), to control the number of active servers for minimizing the data center operational cost (defined as a weighted sum of electricity cost and delay cost) while satisfying carbon neutrality without requiring long-term future information. Unlike prior research on carbon neutrality, we explore demand-responsive electricity price enabled by the emerging smart grid technology and demonstrate that it can be incorporated in data center operation to reduce the operational cost. Leveraging the Lyapunov optimization technique, we prove that CNDC achieves a close-to-minimum operational cost compared to the optimal algorithm with future information, while bounding the potential violation of carbon neutrality, in an almost arbitrarily random environment. We also perform trace-based simulation as well as experiment studies to complement the analysis. The results show that CNDC reduces the cost by more than 20% (compared to state-of-the-art prediction-based algorithm) while resulting in a smaller carbon footprint. Moreover, by incorporating demand-response electricity prices, CNDC can further decrease the average cost by approximately 2.5%, translating into hundreds of thousands of dollars per year.