Can cloud computing be labeled as “green”? Insights under an environmental accounting perspective

The importance of information and communication technology (ICT) sector on the global energy consumption and CO2 emissions tend to grow. ICTs have a fundamental role in collaborating for a sustainable development by providing services in an efficient way, however, its own structure should also follow sustainable principles, for instance, by consuming lower energy amount. The quest for sustainability of ICTs has been focused on data centers (DC) optimization through techniques of sharing infra-structures, which would result in energy efficiency increase, carbon footprint reduction, and reduction of e-waste material disposal. In this scenario, the cloud computing technique rises as the most promising one, often receiving the “green” label. However, this label is usually based on electricity consumption reduction and disregards several other important “green” label-related aspects. This work uses emergy accounting (spelled with an “m”) and direct energy consumption in calculating indicators of eco-energy efficiency for DC operating under traditional and cloud computing techniques. A traditional decentralized DC and a centralized cloud computing DC are herein considered for illustrating figures and for discussion. Results show that centralized DC is able to provide a virtual machine (VM) by demanding 51% less electricity than decentralized DC, and it consumes 87% less electricity to store a byte. Under an emergy accounting perspective, the centralized DC demands 45% less global resources than the decentralized DC to provide a VM while demanding 85% less global resources to store a byte. Although the assessed indicators point out better eco-energy efficiency performance for the DC using cloud computing techniques, labeling it as “green” could be considered as premature due to a lack of threshold which allows for categorizing a system as “green”. Nevertheless, the centralized DC evaluated should be promoted due to its better performance as for the considered indicators.