An Energy Efficient Virtual Machine Placement Scheme for Intelligent Resource Management at Cloud Data Center

In cloud computing environments, virtualization optimally allocates PM (physical machine) resources to users through VM (virtual machines). Operating these PMs leads to significant energy usage, and PM overloading due to high resource demands can breach SLA (Service Level Agreement). Dynamic VM consolidation techniques mitigate this by migrating VMs to enhance resource usage and reduce energy consumption. However, excessive migrations can hamper application performance due to runtime overhead. Effective resource management is crucial to improve cloud operational efficiency. An EE-VMP (Energy Efficient VM Placement) model is proposed to tackle these challenges. It utilizes an evolutionary NN (Neural Network) to predict VM resource needs, optimized by the Stochastic Gradient Descent with Momentum algorithm. Optimal VM allocations are based on these predictions. Experiments using the GCD (Google Cluster Data Set) highlight the model’s efficacy. Compared to existing methods, the proposed model exhibits substantial reductions in power consumption by up to 56.89%, the number of active servers by 37%, and improvement in resource utilization by up to 64.03%.