云计算环境下资源负载均衡调度优化仿真

摘要

云计算环境中,部分节点超载会严重影响集群性能,甚至影响用户体验.为了提高用户服务质量与云计算的整体效率,利用自适应负载均衡对云计算节点资源进行整体优化来平衡各物理节点的计算资源.综合考虑节点的CPU、网络带宽、内存资源、磁盘存储资源四维负载情况,利用新型加权因子方法并结合改进的遗传算法,提出一种多维度资源负载均衡策略,旨在达到服务器负载均衡,同时提高云计算的整体性能.上述方法先通过实时采集各物理节点和虚拟节点的资源信息,利用新改进的加权因子算法和改进的遗传算法结合,迭代求出最佳染色体的负载度、适应度及选择概率.经过模拟筛选后以负载度最小、适应度最强、选择概率最佳的染色体组做为云计算集群环境的节点进行迁移部署.当部分节点的动态负载持续过重时,重新按上述机制计算最佳染色体部署云计算环境,并利用实时迁移命令触发迁移.实验结果表明,在物理机严重超载的情况下,上述方法能在很短的时问内提高负载均衡度并提高各物理机的效率,能够改善集群性能.%In order to improve quality of service and global efficiency of cloud computing,a multi-dimensional resource load balancing method is proposed.The resource information of each physical node and virtual node is collected in real-time,and improved genetic algorithm and improved weighted factor algorithm are used to iteratively obtain the load degree,fitness and selection probability of optimal chromosome.After the virtual screening,the chromosome set with the minimum load,the strongest fitness and the best selection probability is used as the node of cloud computing clustered environment for migration and deployment.When the dynamic load of some nodes is overloaded continually,the optimal chromosome is calculated and cloud computing environment is deployed.In addition,the real-time migration command is used to trigger migration.Simulation results show that when physical machine is overloaded seriously,the proposed method can improve the load balance degree and the efficiency of each physical machine in a short time.Moreover,the clustering performance can be improved.