LRDSM- A Projected Energy Efficient Dynamic Virtual Machine Allocation Framework for Cloud Infrastructure
Nilamadhab Mishra, Ram Kumar, Jyoti Batra, Monica Sankat · 2025
Cloud computing is an alternative advancement allowing us to deploy our entire computer infrastructure online, including hardware and software applications. This study presents a dynamic virtual machine allocation and migration strategy to enhance energy efficiency while preserving cloud data centers' stipulated quality service requirements. Our proposed framework, called LRDSM, seeks to predict the brief CPU utilization of hosts based on the historical usage of data. We use this assessment to identify overloaded and underloaded hosts during live migration. If a host is overloaded, some virtual machines (VMs) are moved to alternative hosts to prevent SLA breaches; conversely, if a host is underloaded, all VMs on that host are attempted to be transferred to alternate machines to facilitate the host's power down. We conducted comprehensive simulation tests using CloudSim 3.0.3, extending many core classes, to assess the efficiency and efficacy of our proposed strategy. Our simulation results demonstrate that the LRDSM approach, utilizing the GCV smoothing parameter, is feasible and may significantly cut power usage in cloud systems.