TenantCPUMan:基于负载分析的多租户动态CPU资源调整

TENANTCPUMAN:MULTI-TENANT CPU DYNAMICAL ADJUSTMENT BASED ON LOAD ANALYSIS

在云计算多租户应用模式中,由于动态的负载变化会影响租户资源分配,可能会出现因负载过高而产生资源竞争或侵占,导致服务质量下降。针对关键的CPU资源,设计一种基于负载分析的多租户应用动态CPU资源调整机制及其原型系统,在租户应用负载信息采集的基础上,运用多元线性回归法对资源分析,根据负载的变化动态调整资源分配,实施租户虚拟机调整、CPU资源管理等。测试结果表明该系统能有效实现多租户资源调整,能应对租户的CPU和内存负荷过载问题,有助于保障租户服务质量。

In the applications of cloud computing and multi-tenant, the dynamic load changes will affect the allocation of tenants. Due to the load is too high,resource competition or occupationmay occur,leading to the decrease of the quality of the service. In view of the key CPU resources, a multi-tenant application dynamic CPU resource adjustment mechanism based on load analysis is designed, and its prototype system has been implemented. It uses the multiple linear regression method to analyze the resource. According to the variation of the load, the allocation of resources will be adjusted, the virtual machine adjustment and the CPU resource management will he implemented. The experimental results show that the system can effectively realize multi-tenant resource adjustment, as well as deal with the problems of the overload of CPU and memory,which is helpful to ensure the quality of service for the tenants.