一种基于负载均衡的云资源调度方法

摘要

Virtualization technology can achieve the distribution of cloud resources according to need.However,it also may cause the heavy load for some physical servers which make system performance reduced.Load balancing of cloud system is the core factor that determines the system computing and service capability.Based on the formal description of cloud resource problem,load balancing coefficient has been defined and load balancing scheduling model for cloud resource has been built.Then,aiming at differences issues such as user demands and servers properties involving in cloud resource scheduling,a method has been proposed focusing on the principle of maximum demand prior and the freest server selected.It has been implemented for algorithm design and analysis further.Dynamic calculation of load balancing coefficient has been put forward as well.The result of simulation experiment has proved that the method proposed has excellent performance for load balancing and wide adaptability especially comparing to the random scheduling and rotation scheduling algorithm.It has also shown that the method can achieve load balancing of cloud resources with reasonably coordinating user demands and the server properties.%虚拟化技术可以实现云系统中资源的按需分配,但同时可能会造成部分物理服务器负载过重,从而导致系统性能降低.云系统负载均衡,是决定系统计算和服务能力的核心因素.在对云资源调度问题的形式化描述基础上,定义了负载均衡系数,构建了云资源调度负载均衡模型.针对云资源调度中的用户需求及服务器属性的差异性问题,提出了一种基于最大需求优先和最闲服务器被选原则的云计算资源调度方法,并进行了算法设计与分析,给出了系统负载均衡系数的动态计算公式.仿真实验结果表明,所提出的方法可得到理想的云计算负载均衡结果,与随机调度、轮转调度等传统算法相比具有优越的负载均衡性能和广泛的适应性.与此同时,所提出的方法可合理协调用户需求和服务器属性,以实现云资源负载均衡.