Cloud Computing-Integrated Load Balancing System for Scalable Web Services

The paper introduces a Cloud Computing Integrated Load Balancing System of Scalable Web Services that helps to address the issues of dynamic workload management, latency and scalability of distributed environment. The suggested framework incorporates the Adaptive Kalman Filter tool to be used as a preprocessor to remove noise and create smooth workload data streams, whereas mRMR (Minimum Redundancy Maximum Relevance) is used as an efficient tool to select the features and improve the model accuracy and lower the computational complexity. A Graph Attention Network is used as the fundamental classifier to model smartly the node interactions and dynamically allocate loads among cloud resources. This system is deployed and tracked on the Databricks Lakehouse platform with MLflow, which allows performing the deployment and tracking the performance in a scaled way. The experimental findings show that the throughput can be significantly improved, and the latency can be reduced and the energy consumption can be decreased with better results than in the traditional methods. The suggested framework provides a basis of intelligent, flexible, and power-efficient load balancing in contemporary cloud-based web service frameworks.