Green Data Centers: A State-of-the-Art on Energy Consumption Estimation and Optimization

Abstract

The rapid growth of digital services and cloud computing has led to a major increase in the number of data centers worldwide. This expansion raises two main challenges for green data center managers: ensuring quality of service in terms of availability, security, and performance, and improving energy efficiency while reducing environmental impact. A data center brings together computing resources, cooling systems, and power distribution. Many research efforts have focused on reducing energy use to improve Power Usage Effectiveness (PUE) while still respecting Service Level Agreements (SLA). Energy optimization can be addressed at the software level through consolidation and workload management, and at the hardware level through well-dimensioned configurations, low-carbon energy sources, and efficient cooling systems. Through a detailed study of the essential concepts of Green Data Centers, this work demonstrates the importance of accuracy in power consumption calculation and the role of optimization achieved through different approaches. The carefully analyzed state of the art reveals both the strengths and the limitations of existing research on Green Data Centers, while also highlighting the specific challenges that remain to be addressed. This thesis highlights the importance of accurate evaluation and optimization of power consumption, both at the level of IT equipment and the whole datacenter, in order to maintain energy efficiency and reduce the carbon footprint. These aspects are considered essential in the context of the continuous growth in demand for digital services and the increasing reliance on datacenters.