一种基于集群的动态负载均衡算法研究

Research on a dynamic load balancing algorithm based on cluster

随着互联网的飞速发展,用户访问量正日益增大,巨大的用户访问量对服务器的性能提出了严峻的考验。在这样的背景下服务器集群出现了,它将很多台服务器集中起来一起进行服务,从而减轻单台服务器的压力。然而在集群中,如何根据服务器性能合理分配任务、避免负载资源浪费是一个关键问题。针对上述问题,文中提出了一种新的负载均衡算法,该算法可以动态更新服务器各项性能指标权重与服务器权值。实验结果表明,该算法在保证单个任务处理时间小于25 ms的同时还能使吞吐率保持在40任务数/秒以上。通过与传统负载均衡算法进行对比,可以看出文中算法能更好地实现集群的负载均衡。

With the rapid development of the Internet,the amount of the users is increasing day by day,which puts forward a severe test on the performance of the server.Under this background,a cluster of servers appeared,it can make a lot of servers provide service together,so as to reduce the pressure of a single server.However,under the cluster,how to allocate tasks reasonably according to the server performance and avoid the waste of load resources is a key problem.Aiming at the above problems,this paper proposed a new load balancing algorithm,which can dynamically update the weights of each server performance index and server weight.It was proved with an experiment thatthe proposed algorithm can ensure the processing time of a single task can be less than 25 ms and the throughput can be maintained above 40 tasks per second.After comparing with the traditional load balancing algorithm,it can be known that the new algorithm in this paper can better realize the load balancing of the cluster.