Power and Carbon Footprint Evaluation and Optimization in Transitioning Data Centres

Abstract

Data centers are among the largest consumers of energy and significant contributors to global carbon emissions. This paper presents a comparative evaluation of two mathematical energy consumption models—the linear and cubic models—within a simulated cloud data center environment using the CloudSim toolkit. Key performance indicators were analyzed, including total energy consumption, carbon dioxide (CO₂) emissions, cost, power usage effectiveness (PUE), carbon usage effectiveness (CUE), response time (latency), and packet loss, under varying workload conditions.The cubic model demonstrated greater sensitivity to workload fluctuations and more accurately captured the dynamics of energy consumption in high-utilization environments, while the linear model offered computational simplicity and conservative estimates. The integration of 50% renewable energy sources led to substantial reductions in both emissions and operational costs. Specifically, the linear model recorded a 33.3% reduction in CO₂ emissions, a 71.5% improvement in CUE, and a 15.4% decrease in cost, while the cubic model achieved a 33.3% reduction in CO₂ emissions, a 23.2% improvement in CUE, and a 24.6% decrease in operational cost—demonstrating the effectiveness of clean energy adoption without compromising system performance or service quality.