Optimized Virtual Machine Scheduling for Green Cloud Computing Using MLP-LSTM with Sheep Flock Optimization

With the growing demand for cloud computing services, it is important to efficiently manage resources using proper strategies in order to save energy consumption while retaining high-level performance. The innovative framework proposed here utilizes the MLP and LSTM models with a bio-inspired optimization algorithm in order to improve workload predictions and resource allocation for Virtual Machine (VM) scheduling. This work has an effective framework for energy-aware virtual machine scheduling in cloud computing environments using MLP-LSTM hybrid models optimized by the Sheep Flock Optimization algorithm, and comprehensive datasets it uses involve essential workload features, such as CPU and memory utilization. The proposed methodology utilizes the min-max normalization during data preprocessing, which ensures uniform scaling of all features. In the integrated MLP-LSTM model, an impressive accuracy of 98.7% was obtained for the task of prediction. Moreover, it was even better than other existing methodologies including Random Forest and Decision Tree. Using the predictions from the MLP-LSTM model, the SFO algorithm optimizes VM placement and migration, reduces energy consumption by a large margin, but at the same time maintains balanced resource distribution across all the physical servers. The results demonstrate the enhancement of both operational efficiency and energy usage, hence the sustainability of cloud computing. The implementation of the framework in Python gives it a realistic solution to situations that pose challenges to resource management in modern cloud infrastructures. This research validates the potential use of an advanced ML approach in conjunction with optimization techniques that can succeed in creating a green and efficient overall ecosystem for cloud computing.