Adaptive Multi-Objective Virtual Machine Consolidation for Energy-Efficient Cloud Data Centers

Cloud data centers (CDCs) widely adopt virtual machine (VM) consolidation strategies to optimize resource utilization and reduce energy consumption. Existing strategies assume that resource utilization doesn’t affect VM performance. They fail to consider that excessive resource utilization can negatively impact performance, increasing Service Level Agreement (SLA) violations. Furthermore, the current methods also utilize historical or predicted utilization-based algorithms to identify overutilized hosts. This can cause high SLA violations, unnecessary VM migrations, and high energy usage, all of which lower the overall performance of CDCs. This research proposes an adaptive multi-objective virtual machine consolidation (AMOVMC) strategy that integrates future workload predictions and past resource utilization trends. This approach prevents VM performance degradation and improves the quality of services (QoS) using a balanced load approach. AMOVMC uses a dynamic multi-weight host detection policy and an adaptive VM placement technique to balance SLAV and energy use while reducing the number of unnecessary VM migrations. Extensive simulations using real-world datasets from PlanetLab, Google Cluster, and Alibaba Traces demonstrate the effectiveness of our approach. This approach reduces energy consumption by 8.19%, VM migrations by over 7.93%, and ESV by over 67.09% in the PlanetLab Trace compared to the best-performing state-of-the-art VM consolidation approach. Similarly, in the Google Cluster and Alibaba Traces, AMOVMC achieves energy savings of up to 9.35%, reduces VM migrations by 8.5%, and decreases ESV by over 80.3%. The results confirm that AMOVMC provides a highly effective solution for VM consolidation. It enhances overall CDC performance while ensuring QoS.