Intelligentmulti-Objective SLA-Aware Load Balancing Framework for Cloud Task Scheduling Using GWO

Abstract

The Cloud Computing (CC) environment is dynamic in nature, and workloads keep changing between the distributed resources. This kind of fluctuation can result in overloading of Virtual Machines (VMs), high response times, and breach of Service Level Agreements (SLAs). To address this issue, this paper proposes an intelligent SLA-aware framework for adaptive load balancing based on the Grey Wolf Optimizer (GWO). The scheduling issue is formulated as a multi-objective optimization problem, which allows the optimal redistribution of workload based on current service requirements to enhance the overall performance and meet the requirements of the SLA. Clouds were used to test the framework in a simulated environment where $\mathbf{n}$-tasks, $\mathbf{5 0}$-VMs and $\mathbf{n}$-brokers control multi-cloud traffic. The performance measures, such as the makespan, the throughput, resource utilization, and the compliance with SLA, were used to evaluate the candidate schedules. In contrast to Round Robin (RR), which applies tasks to jobs in a cyclic manner and does not consider system qualities, and First-Come-First-Served (FCFS), which allocates tasks in a strictly sequential manner and does not consider the load of the system, the GWO-based scheduler dynamically distributes workloads to avoid congestion and worsening of performance. It is observed that, of 2,501 executed cloudlets, 2,207 reached the SLA deadline of 80 -time units, an SLA satisfaction rate of 88.24% and a violation rate of 11.76%, and this maximizes efficiency, fairness, and system reliability.