基于文化算法的负载均衡自适应机制

负载均衡是解决Web集群系统容量和伸缩能力的重要方法,但通常使用的单纯的加权轮叫调度算法依然会导致服务器间的负载不平衡。本文分析了影响Web服务器性能的主要因素,提出了一种负载均衡的自适应机制。该方法将文化算法(CultureAlgorithmsCA)应用到对服务器性能权值的进化计算中,通过评价服务器的负载状况,获得优化的性能权值,并自适应地转换到集群的分配器中,使事务在集群系统中得到合理分配。模拟实验证明,随着访问量的逐渐增多,每台服务器都趋近于最佳负载,系统达到好的使用效果。

A Self-similarity Load Balancing Strategy Based on Culture Algorithm

Load balancing is an important method to solve the capacity and scalability of web server system,while weighted round-robin scheduling leads occasionally to unbalancing.This paper analyses the main factors which effect the performance of web server,and proposes a new strategy,which applies culture algorithm into the evolutionary calculation for the weighted values of server.This method evaluates the loading status of the severs firstly to get optimized weighted value,then puts them into balancing distribute program as self-similarity strategy,which guarantees the tasks distributed reasonably.Simulation experiments show that every server tends to reach good balancing effect as the count of access raising,and the system will run in high efficiency.