A Study of Competitive Cloud Resource Pricing under a Smart Grid Environment

Abstract

In the current IaaS cloud market, to achieve profit maximization, multiple cloud providers compete non-cooperatively by offering diverse price rates. At the same time, tenant consumers judiciously adjust demands accordingly, which in turn affects cloud resource prices. In this paper, we tackle this fundamental but daunting cloud price competition problem with Bertrand game modeling, and propose a dynamic game to achieve Nash equilibrium in a distributed manner. Specifically, we realistically consider spot electricity prices under a smart grid environment, and systematically investigate the impact of different system parameters such as network delay, renewable availability, and cloud resource substitutability. We also perform stability analysis to investigate the convergence of the proposed dynamic game to Nash equilibrium. Cooperation among cloud providers can achieve aggregate cloud profit maximization, but is subject to strategic manipulations. We then propose our Striker strategy to stimulate cooperation, the efficiency of which is validated by repeated game analysis. Our evaluation is augmented with realistic electricity prices in the spot energy market, and reveals insightful observations for both theoretic analysis and practical pricing scheme design.