Large-scale Distributed Systems and Energy Efficiency: A Holistic View

Chapter 5Power modeling

Jason Mair,¹ Zhiyi Huang,¹ David Eyers,¹ Leandro Cupertino,² Georges Da Costa,² Jean-Marc Pierson,² and Helmut Hlavacs³

¹Department of Computer Science, University of Otago, Dunedin, New Zealand

²Institute for Research in Informatics of Toulouse (IRIT), University of Toulouse III, Toulouse, France

³Faculty of Computer Science, University of Vienna, Vienna, Austria

5.1 Introduction

Power consumption has long been a concern for portable consumer electronics, with many manufacturers explicitly seeking to maximize battery life in order to improve the usability of devices such as laptops and smart phones. However, it has recently become a concern in the domain of much larger, more power hungry systems such as servers, clusters, and data centers. This new drive to improve energy efficiency is in part due to the increasing deployment of large-scale systems in more businesses and industries, which have two primary motives for saving energy. Firstly, there is the traditional economic incentive for a business to reduce their operating costs, where the cost of powering and cooling a large data center can be on the order of millions of dollars [1]. Reducing the total cost of ownership for servers could help to stimulate further deployments. As servers become more affordable, deployments will increase in businesses where concerns over lifetime costs previously prevented adoption. The second motivating factor is the increasing awareness of the environmental impact—greenhouse ...