Performance Evaluation of Dynamic Load Balancing Algorithms

Abstract

Efficient task scheduling mechanism can better meet the users’ QoS requirement, and achieve the load balancing in physical hosts, so the cluster system with high scalability and reliability can effectively improve the information utilization of the system, thereby enhance the overall performance of the Web server cluster system. In order to build a network service system with better scalability and reliability, this paper describes the round-robin scheduling algorithm of LVS cluster system, least-connection scheduling algorithm, weighted least-connection scheduling algorithm and a prior proposed new weighted value assigned scheduling algorithm, dynamic adaptive feedback load balancing strategy. Meanwhile, it takes simulation experiment for the round-robin scheduling algorithm of LVS, least-connection scheduling algorithm, weighted least-connection scheduling algorithm and the new weighted value assigned scheduling algorithm, dynamic adaptive feedback load balancing strategy and take comparative analysis for the experimental data, through the analysis and assessment, effectively point out the advantages and disadvantages of the existing load balancing strategy. It is conducive to better improve the existing equalization algorithm performance deficiencies and propose optimum load balancing strategy

DOI : http://dx.doi.org/10.11591/telkomnika.v12i4.4731