Energy efficiency of data center operating practices: Server clustering, powering on/off, and bang–bang control

Abstract

Data centers today account for up to 3% of all global electricity consumption and produce 200 million metric tons of CO 2 . Hence, improving energy efficiency is an important operational concern. This article examines common operating practices such as server clustering, powering on/off, and bang–bang control in terms of energy efficiency. Extending the mixed integer programming (MIP) formulation in a previous research study, we introduce new constraints reflecting the operating practices adopted for operational convenience. We develop an algorithmic strategy for constructing energy efficient clusters of servers using two upper bounds on the cluster size. Numerical experiments show that server clustering and bang–bang control do not diminish energy efficiency, and that powering on/off by itself is insufficient. Our results produce a conclusion that differs from the published literature on data center design and operations. © 2017 Wiley Periodicals, Inc. NETWORKS, Vol. 71(2), 107–119 2018

Online Power-Aware Deployment and Load Distribution Optimization for Application Server Clusters

“…In the solution method for the cluster optimization problem, [19], [26]- [28] use heuristic or greedy algorithms on the whole. Reference [8] first reduces the number of variables by half through a theoretical analysis and then uses a binary search; [5] uses Generalized Benders Decomposition; [18] and [16] use GLPK and CPLEX, respectively; and [14] proposes a bisection method and several heuristic methods. However, these studies do not evaluate the efficiency of their solution methods.…”

“…The Flower Pollination Algorithm (FPA) is a populationbased optimization algorithm that was proposed in 2012 by Algorithm 1 Correction Operator 1: //for (15), (16), and (17) 2: if(n i,1 , l i,1 , n i,2 , l i,2 , or n i,1 + n i,2 is smaller/bigger than its lower/upper bound (i ∈{0,. . .…”

“…This is dangerous because even a minor load burst may make the server unresponsive. Therefore, the server's utilization must stay within a reasonable range [16].…”

A QoS-aware workload routing and server speed scaling policy for energy-efficient data centers: a robust queueing theoretic approach

“…Xiong et al [25] formulated an optimization model for server clustering and proposed a meta-heuristics algorithm to solve the mixed integer non-linear programming. Urgaonkar et al [20], Arrubla et al [12], and Cho and Ko [9] also belong to this category. Under the setting by planning, research at the operation level studied the optimal load balancing and server speed scaling to process real-time requests.…”

Preface: Recent advances in telecommunications networks planning and operation

“…Energy efficiency is therefore an important problem to be addressed. Cho and Ko develop an algorithmic strategy for clustering servers in an energy efficient way, and provide new insights into the use of clustering, powering on/off and bang‐bang control operational strategies.…”