Energy efficiency in colocation data centers: A joint incentive mechanism approach

Abstract

Colocation data centers (or colocations, for short) are important participants in emergency demand response (EDR) programs. One key challenge in colocations is that tenants control their own servers, thus, may not coordinate to reduce their power consumption. In this paper, we propose a joint truthful incentive mechanism Co-Colo to encourage tenants joining EDR programs, which includes a local optimization mechanism (LocalOpt) and a global optimization mechanism (GlobalOpt). In LocalOpt, tenants are motivated to improve the energy efficiency locally. In GlobalOpt, tenants can request some public server resources to improve the energy efficiency. By jointly considering the two mechanisms, Co-Colo effectively reduces the energy-saving cost. A (1 + ε)-approximation algorithm is proposed to obtain the asymptotic optimal energy-saving scheme. We also consider a special case when the public resources are sufficient, and design a 2-approximation algorithm. Furthermore, the robustness of the proposed algorithms are proved. Trace-driven simulations verify the effectiveness and feasibility of Co-Colo.