Towards Improved Datacenter Assessment: Review and Framework Proposition

Our growing needs for social networks, cloud storage and more recently machine learning have fueled the increasing demand for datacenters (DC). It is estimated that by 2030, in the US alone, datacenter power consumption could more than double from 2022 [1]. In Europe, the energy consumption is expected to rise by 28%, from 77TWh to 99TWh [2]. This surge, coupled with the increasing scrutiny imposed on Information and Communication Technology (ICT) from stakeholders, regulators and competition, regarding environmental impacts, and to stay on the path of net-zero, has spotlighted datacenters as key contributors to these environmental concerns. Consequently, companies have set major milestones for the next decades in terms of renewables, energy consumption and water use. This paper aims to shed light on the imperative necessity of revisiting and evaluating various metrics and methodologies used to gauge the impact of datacenters, extending beyond merely assessing sustainability factors. More in detail, we focus on four paramount criteria which encompass the whole datacenter lifecycle and its direct and indirect impacts: environmental impact, economic performance, ecosystem integration and external influence. These are usually evaluated through three types of analysis: single indicators, lifecycle analysis and multi-criteria assessment, all of which are analyzed here.