Propagating Electricity Bill onto Cloud Tenants: Using a Novel Pricing Mechanism

Data centers spend millions of dollars on their electricity bills annually. Therefore, there is an interest among data center operators to control the electricity usage so as to minimize energy expenditure. However, when it comes to cloud data centers, the electricity usage is mainly controlled by the tenants. Yet, since most cloud data centers charge their tenants with flat rates, the tenants do not have incentives to change their electricity usage to contribute in cutting electricity bills by participation in demand response. Accordingly, in this paper, we propose a game-theoretic framework together with a time varying pricing (TVP) mechanism for cloud data centers to charge their tenants. In this approach, the TVP propagates the actual energy bill, which comprises both demand and energy charges, onto tenants' service costs. As an extension to this core idea, the time- varying amount of renewable energy generated by data center's on-site renewable generators is also taken into account to affect the payments. Our proposed pricing method is evaluated under various experimental data and simulations. We show that TVP can boost data center's profit by 8.2% and reduce the energy bill by 33.0% and improve tenants' aggregate surplus by 12.3%, when comparing it with a flat rates model that uses a widely employed billing method in today's cloud data centers.