Energy-Aware Virtual Machine Scheduling for Sustainable Cloud Data Centers

Abstract

This can be attributed to the fact that the fast growth of cloud computing services has caused a sharp rise in energy consumption in large-scale data centers increasing both the cost of operations and the environmental issue. Cloud infrastructure that is sustainable requires resource management techniques that reduce the energy consumption but ensure Quality of Service (QoS). The present paper is a proposal of an energy-conscious virtual machine (VM) scheduling architecture aimed at maximizing the power used by cloud data centers by dynamically consolidating workloads and allotting resources. The model proposed combines the energy consumption modeling with a multi-objective model which takes into account the CPU utilization, VM migration cost, SLA violation rate, and thermal constraints. It includes a predictive workload analysis mechanism to predict resource demand and to be able to place VM proactively and scale dynamically. The scheduling algorithm has taken advantage of a hybrid metaheuristic approach that balances the server usage and minimizes the amount of idle power it uses. The experimental assessment of a simulated cloud system shows that the offered method enables to save substantially on the total energy consumption and carbon footprint than the standard First-Come-First-Serve (FCFS) and the static threshold-based strategies of scheduling. Findings point at the improvement of the energy efficiency, the diminished SLA violations, and the general stability of the system. The results point to the efficiency of smart, energy-sensitive scheduling tools in the context of encouraging green and economically efficient cloud-based data center activities.