Adaptive Load Balancing Algorithm Based on Prediction Model in Cloud Computing

Abstract

In cloud computing, load balancing is required to distribute the dynamic local workload evenly across all the nodes. It helps to achieve a high user satisfaction and resource utilization by ensuring an efficient and fair allocation of every computing resource. Although many load balancing schemes have been presented in Cloud computing, there is no scheme providing the elasticity and adaptive adjustment in cloud computing. In this paper, an Adaptive Load Balancing Algorithm based on load prediction model (ALBA) was proposed to improve the resource utilization. When the load in the cluster of virtual machines is lower than the minimal threshold, the ALBA scheme can callback the resources of the cluster. While the load in the cluster of virtual machines is higher than the maximum threshold, the ALBA will adaptively add new virtual machines to balance the computation load and ensure the response time. To avoid the data fluctuation causing by the real-time load acquisition, a load prediction model was introduced and used to improve the accuracy of load prediction. The extensive experiments with CloudSim demonstrate that the proposed adaptive load balancing algorithm -- ALBA, can improve the resource utilization as well as reduce the respond time of tasks.