云计算环境下基于非均匀窗口蚁群行为的负载平衡算法

Load balancing algorithm based on non-uniform window ant colony behavior in cloud computing environment

针对云计算环境下可能面临请求过载和较长响应时间的问题,在非均匀窗口蚁群行为思想的启发下,提出一种负载平衡算法。假设云环境下任何时候虚拟机都处于过载状态,即服务提供者不得不分配资源。根据该假设,对可利用的资源合理优化,优化过程中动态代理和静态代理同时进行,其中蚁群行为被用于负载平衡,通过加载资源到所有的虚拟机上来实现系统平衡。利用Cloud Sim仿真器模拟云计算环境进行实验分析,实验结果表明,与容错分簇的负载均衡感知(tolerant cluster-aware,TCLB)、基于博弈论的负载均衡算法(scheduling strategy on load balancing,SSLB)和基于蜜蜂行为的负载均衡算法(honey bee behavior inspired load balancing,HBB-LB)相比,提出的算法分别节省了37%,8%和4%的响应时间,最大完成时间也大幅度降低,整体性能有所提高。

In view of request overloading and high response time in cloud computing environments, inspired by the idea of the non-uniform window ant colony behavior, a new load balancing algorithm is proposed. We introduce a working hypothesis that any virtual machine is under the condition of overloading, that is, the service provider has to allocate resources.According to the assumption, the available resources are optimized, where the dynamic and static agents are processed at the same time, and the ant colony behavior is used for load balance. System balance is achieved by means of loading resources into all virtual machines. The CloudSim simulation is used to emulator cloud environment for experiment analysis.Experimental results show that compared with load balancing of fault-tolerant cluster-aware ( TCLB) , scheduling strateg}^ on load balancing( SSLB) and honey bee behavior inspired load balancing( H BB-LB) ,the responsible time of the proposed algorithm is less than that of TCLB, SSLB and HBB-LB by 37% , 8 % and 4 % respectively. And the maximum completion time is also reduced significantly with overall improved performance.