基于排队论综合指标评价的动态负载均衡算法.PDF

研究与开发 基于排队论综合指标评估的动态负载均衡算法 王文博，叶庆卫，周宇，陆志华 （宁波大学信息科学与工程学院，浙江宁波 315211 ） 摘 要：互联网通信、计算机集群和云环境均具有一定的复杂性和动态性，极易发生负载失衡，从而降低服 务效率、增加能耗。因此，负载均衡技术成为重点研究课题。现有的负载均衡策略均是以 CPU、内存、进程 等参数的占用率来评估服务器当前的负载情况，但服务器负载情况的复杂性往往使其难以得到准确评估。针 对该问题，提出了一种基于排队论综合指标评估的动态负载均衡算法，首先引入排队论模型评估各服务器的 实时负载情况，然后根据各服务器的负载综合指标，将输入队列中的任务逐一分配给各服务器。实验结果表 明，该方法可有效平衡各服务器的负载且减少任务请求的平均等待时间。 关键词：负载均衡；排队论；性能 中图分类号：TP393 文献标识码：A doi: 10.11959/j.issn.1000−0801.2018204 Dynamic load balancing algorithm based on queuing theory comprehensive index evaluation WANG Wenbo, YE Qingwei, ZHOU Yu, LU Zhihua School of Information Science and Engineering, Ningbo University, Ningbo 315211, China Abstract: Internet communication, computer cluster and cloud environment have complex and dynamic characteris- tics, which can cause load imbalance easily, reduce the service efficiency and increase the energy consumption. Therefore, the load balancing technology becomes the focus of research. The existing load balancing strategy uses the occupancy of CPU, memories, processes to estimate the current load of each server. But it is hard to guarantee its ac- curacy. Aiming at this problem, a dynamic load balancing algorithm based on queuing theory comprehensive index evaluation was proposed. Firstly, queuing theory model was introduced to estimate the real-time load of each server, and then the tasks of input queue was distributed to each server separately according to the load comprehensive index of each server. Experimental results show that this method can balance the load of each server effectively and reduce the average waiting time of the task requests, which is of great application value. Key words: load balan