基于GM(1,1)预测和虚拟机迁移的云计算负载均衡策略设计

Design for Load Balance Strategy of Cloud Computing Based on GM(1,1) Prediction and Virtual machine Migration

摘要:

针对云计算中心虚拟机集群负载的不均衡问题,设计了一种基于GM(1,1)预测和虚拟机迁移的负载均衡策略；首先,描述了云计算中心的负载均衡原理,介绍了经典ELB算法并总结了其缺点,然后,设计了基于GM(1,1)的虚拟机负载预测算法,能根据虚拟机的历史负载信息来预测下一时刻的负载信息,能有效克服ELB算法中仅依赖当前时刻负载而进行虚拟机的删除和增加,同时通过设定不同的阈值来实现虚拟机的增加、删除和迁移,最后,定义了基于预测机制和虚拟机迁移的负载均衡算法,能在创建虚拟机时根据用户的请求定制AMIs模板,从而提交用户请求的响应速度；在CloudSim环境下进行实验,实验结果表明文中方法能有效地实现云计算中心虚拟机集群的负载均衡,与其它方法相比,具有负载均衡高和负载均衡效率高的优点,是一种有效的云计算环境的负载均衡策略。

Abstract:

Aiming at the load unbalance problem of cloud computing virtual machine, a load balance strategy based on GM(1,1) prediction and virtual machine migration is proposed. Firstly, the load balance principle is described and the classic ELB algorithm is introduced, and the defects is summarized. Then the load balance algorithm based on GM(1,1) is designed, the load information can be predicted based on history load information. It can realize add and delete of the virtual machines relying on the current load of ELB algorithm. The threshold can be set as different values for add, delete and migration. Finally, the load balance algorithm based on prediction mechanism and virtual machine migration is defined. The experiment is operated in the CloudSim, the experiment shows the method in this paper can realize the load balance in cloud computing virtual machine, and compared with the other methods, it has high load balance and load balance efficiency, therefore, it is an effective load balance strategy in cloud computing.