EnergIT: A Methodology for the Incremental Green Design of Data Centers

Introduction

Research on Green ICT has been active for over a decade and is currently structured as a multi-disciplinary research area offering consolidated end-to-end methodologies (cf. Schultz, 2009; Patel et al., 2002; Bash & Foreman, 2007; Shah et al., 2008; Brown & Reams, 2010) and numerous in-depth systematic studies (cf. Li et al., 2007; Patel et al., 2002). Within the Green ICT area, the green data center research stream is particularly active, with a correspondingly broad array of market solutions. Hardware vendors have included energy efficiency among their base set of target parameters, and are constantly benchmarked against their performance-to-power ratio (cf. www.vmware.com). Whether accounting for these benefits when assessing the payoffs of green is questionable. Large green projects are often justified on the basis of all possible economic returns, including the cost reductions from virtualization. However, part of these benefits can be obtained by applying virtualization to legacy servers without investing in new expensive green hardware, such as high-performance blade servers and enclosures or dedicated precision cooling infrastructures. In this case, addressing small data centers needs means assessing the gap between these benefits and the maximum possible returns involving investments in the newest and most efficient hardware.