LBGWO: A Load Balancing Based Strategy for Large Scale Task Scheduling in Cloud Computing

Abstract

Efficient task scheduling and load balancing are challenging in heterogeneous cloud computing environments due to uneven workload distribution and increasing computational demands. This paper proposes LBGWO (Load-Balanced Grey Wolf Optimization) which integrates an adaptive load-balancing strategy with Grey Wolf Optimization. The proposed method finds overloaded VMs using an adaptive threshold based on mean load and standard deviation and reallocates tasks to the lest-loaded VMs. Experiments are conducted using Google Traces 2019 workloads of 2000-3500 tasks in CloudSim 3.0. The proposed LBGWO is compared with IPSO, PSO and GWO using makespan, degree of imbalance, response time, throughput and convergence. Results show that LBGWO algorithm consistently achieves superior performance with 13-25% improvement in makespan, 36-49% improvement in degree of imbalance, and 3-18% improvement in response time over its competitor IPSO algorithm. LBGWO also achieves the highest throughput and faster convergence, demonstrating its effectiveness for load-balanced cloud task scheduling.