Dynamic Load Balancing in Cloud Computing A Novel Approach

Abstract

Dynamic load balancing is a critical challenge in cloud computing environments due to heterogeneous resources, fluctuating workloads, and strict Quality of Service (QoS) requirements. Inefficient task allocation across virtual machines (VMs) leads to increased response time, processing time, and makespan, thereby degrading overall system performance. This research proposes a set of dynamic and hybrid load balancing algorithms aimed at optimizing these key performance metrics in multi-tenant cloud infrastructures. newlineThe primary contribution is the Enhanced Model Priority Based Throttled Load Balancing (EMPBT-LB) algorithm, which dynamically assigns tasks based on VM availability, task priority, and current workload conditions. The algorithm incorporates predictive analytics and state-aware monitoring to improve responsiveness and resource utilization in heterogeneous cloud environments. In addition, an Enhanced Shortest Job First with Priority (ESJFP) scheduling algorithm is introduced to minimize makespan by considering both task length and VM processing capability (MIPS). A priority-aware dynamic load balancing model with pre-emption support is also proposed to efficiently manage real-time and high-priority tasks. newlineThe proposed algorithms are evaluated using the CloudAnalyst simulation tool built on CloudSim, under varying workload scenarios and fixed VM configurations. Performance results demonstrate significant reductions in response time, processing time, and makespan, along with improved resource utilization, compared to conventional load balancing techniques. The effectiveness of the proposed methods is further validated through application scenarios in healthcare and e-education systems. newlineOverall, this work provides scalable and efficient load balancing solutions that enhance cloud performance and reliability, meeting the QoS requirements of modern cloud-based applications. newline newline