Enhancing energy efficiency in cloud computing through regression models: A data-driven approach with experimental validation

The rapid growth of Cloud Computing and the expansion of large data centers have led to significant increases in energy consumption for hardware and cooling systems. This surge in power usage has raised operational costs, making energy efficiency a pressing issue for data center management. Virtual Machines (VMs) consolidation is a widely studied strategy to mitigate these costs by minimizing the number of active physical servers while maintaining user Service Level Agreements (SLAs). However, the success of consolidation largely depends on accurately forecasting VM resource needs. This paper presents the design and development of an energy-aware VMs allocation system that leverages predictive regression models to forecast future computational demands (CPU) of each VM. By anticipating these needs, the system optimally allocates VMs to available servers, effectively balancing energy savings with performance. The experimental evaluation, based on regression modeling exploited to forecast CPU and energy usage, shows a comparative analysis among several approaches and demonstrates the effectiveness of machine learning-driven allocations of VMs in optimizing resource utilization and reducing energy consumption.