A three-tier energy efficient architecture integrating virtual machine allocation and consolidation leveraging NSGA-II and LSTM for cloud data center

Inadequate resource utilization of physical machines is a primary cause of high energy consumption in cloud environments. Virtual machine allocation and consolidation play a vital role in efficiently utilizing data center resources. Existing research typically focuses either on optimizing the allocation of virtual machines (VMs) to physical machines (PMs) or on consolidating VMs based solely on current workload information. This paper proposes a three-tier architecture that integrates VM allocation and consolidation by leveraging both current and predicted workloads. By incorporating accurate workload prediction instead of relying only on real-time usage, the proposed method proactively improves placement decisions, thereby minimizing frequent VM migrations caused by dynamic workload changes and reducing the risk of SLA violations. Tier 1 maps VMs to PMs optimally using a Non-dominated Sorting Genetic Algorithm II-based virtual machine allocation and consolidation (NSGAII-VMAC) algorithm. Tier 2 employs a Long Short-Term Memory (LSTM) model to predict VM workloads and compute the predicted workloads of PMs. Finally, Tier 3 performs consolidation in three steps: first, it identifies overloaded and underloaded PMs using current and predicted workloads; next, it selects VMs from overloaded PMs for migration using a VM selection method inspired by the Pareto front strategy; and lastly, it determines the destination PM for placement using NSGAII-VMAC. Extensive experiments demonstrated that the proposed approach reduced energy consumption by up to 50.93%, lowered the number of VM migrations by 30.57%, and reduced the number of active servers by 60.39% compared to baseline methods. For workload prediction, the LSTM model achieved an RMSE as low as 0.028 and an MAE as low as 0.021, outperforming state-of-the-art models. These results validate the effectiveness of a prediction-enhanced framework for energy-efficient cloud data center management. • Proposed a Three-tier architecture for virtual machine allocation and dynamic consolidation. • Application of NSGAII-VMAC for optimization and LSTM for prediction of resource usage. • Proposed improved decision-making using dynamic consolidation resource usage prediction. • Proposed a bi-objective method for VM selection using a Pareto front strategy. • Presented significant reduction in energy consumption, active servers and migration counts.