Enhancing Cloud Sustainability through AI-driven Energy Management and Resource Consolidation

The increasing cloud computing and artificial intelligence workloads have accelerated the need for energy and the emission of greenhouse gases from data centres. This work proposes an adaptive resource consolidation and predictive energy management integrated to adapt the power consumption while maintaining a defined Quality of Service (QoS). The design and implementation of simulation engines that are energy aware and the research that pertains to scheduling workloads that are carbon aware are discussed, followed by an integrated AI infrastructure that combines predictive workload energy demand, carbon intensity, and the dynamic consolidation of virtual machines and containers, which is closed with a reproducible evaluation framework based on well-known simulation tools. The combination of artificial intelligence consolidation and carbon-sensitive workload distribution, which is explored in literature and in the simulation with the defined case studies, acts in line with the stipulations in the service level agreements to reduce energy and carbon emissions significantly. The study closes with a practical exposition of the balance of yielded results, suggested improvements to the work, and prospective investigations, namely, coordination across multiple clouds and remaining resources.