Linear Regression And Recurrent Neural Network Ensemble For Swift Tensile Load Balancer For Static And Dynamic Clouds

Abstract

Efficient load balancing is essential for optimizing resource utilization and performance in both static and dynamic cloud environments. Dynamic cloud environments are used to have frequent workload fluctuations, thus, demand adaptive load balancing strategies are essential to handle the criteria.In general, the load in a cloud computing refers the twin paired loads namely computational load and communication load. Real-time monitoring and machine learning based resource reallocation is the most modern innovative strategy to handle the balance between the computational and communication loads. This work introduced as “Linear Regression and Recurrent Neural Network Ensemble for Swift Tensile Load balancer for Static and Dynamic Clouds (LRESTL)” is the hormonic integration of three novel modules namely Tailored Elastic Net Load Prognosticator, Custom Echo State RNN Load Tracker, and Swift Tensile Consolidator”. LRESTL is implemented and evaluated in a real-time cloud environment to assess benchmark metrics such as resource utilization, balance degree, average response time, migration cost, throughput based on the number of tasks, and throughput based on the number of virtual machines.