EMaC: Dynamic VM Consolidation Framework for Energy-Efficiency and Multi-metric SLA Compliance in Cloud Data Centers

Cloud industry is facing significant challenges related to resource under-utilization and high energy consumption. Virtual Machine Consolidation serves as an effective solution to address these issues. However, Virtual Machine Consolidation is a challenging task because the issues such as dynamic nature of workloads, the complexity of decision-making for resource optimization, the need to meet Service Level Agreements (SLAs), and the risks associated with aggressive consolidation must be addressed properly. This work proposes a multi-objective algorithm named Energy-efficient Multi-objective Virtual Machine Allocation and Consolidation (EMaC) that aims to optimize Virtual Machine Consolidation in a datacenter. The algorithm leverages server clustering approach in order to facilitate optimal Virtual Machine Allocation to streamline energy management and reduce server selection time. The algorithm further adapts to changing workloads and reduces frequent server overloading events by incorporating a dynamic threshold policy. Moreover, the proposed framework integrates SLA awareness into server selection, striking a balance between energy efficiency goals and SLA compliance. Performance of the algorithm is validated against benchmark as well as state-of-the-art policies on Cloudsim simulator by utilizing real workload traces collected from PlanetLab and Bitbrains datacenter. From extensive experimentation, the framework demonstrates encouraging results when evaluated against commonly used performance metrics. The EMaC algorithm shows a maximum reduction of 33.12% and 73.01% in energy consumption and SLA violations respectively with the traces of PlanetLab and 39.68% and 50.63% respective reduction with traces of Bitbrains.