基于云计算的作业管理系统CJMS的负载均衡的研究

Research on Load Balancing of Cloud Computing-Based Job Management System CJMS

作业管理系统常采用集中式负载分散策略实现负载平衡;但是在集中式负载分散策略中,中央控制器容易引起单点故障,成为系统瓶颈;尤其是云计算环境下,服务器数量巨大,中央控制器将承担非常大的负荷;为了提高云计算系统的效率,提出了一种基于分布式分组策略的负载均衡的实现方法,将集群系统中的服务器按照地理位置进行分组,并重点介绍了组内负载分散和组间负载分散的实现过程;该方法通过采用改进的负载信息收集策略和作业转送策略,节省了系统资源,取得了良好的效果。

In general job management system, we always adopt the centralized load diversification strategy for load balancing. But the central controller is the system＇s bottleneck due to its inadequacy of being easy to cause a single point of failure, especially in the cloud computing environment, where there exist a huge number of servers, and the central controller will bear a very large load. Therefore, based on distributed group strategy, proposes a load balancing algorithm which groups the servers in the cluster system by geographical position, and focuses on the load balancing algorithm within the group and between groups. By using an improved load information collection strategy and job transferred strategy, the algorithm saves a lot of system resources and has achieved good results.