摘要
现有业务网云平台存在扩容频繁、资源利用率低等问题,为此提出一种业务网云化准入的资源池容量规划方法。首先,通过对业务进行中长期预测确定业务网资源池总容量需求;然后,以各类规格虚拟机数量最小为目标,参考虚拟机资源约束、资源利用率和负载均衡综合评估整个资源池各个规格虚拟机的数量;最后,以最小化能耗和最小化最大链路为目标,结合多维资源约束,实现物理服务器容量的规划。实验表明,采用本文提出的物理与虚拟资源匹配模型在高请求和进程的情况下仍然具有较高的性能。这个结果有利于提高网络资源利用率,降低业务运营成本。
Aiming at the problems of frequent capacity expansion and low resource utilization of the existing clouding platforms for service networks, a resource pool capacity planning method is proposed for service network clouding access. First, the total capacity demand of the service network resource pool is determined through medium and long-term prediction of the service. Then, with the goal of minimizing the number of virtual machine collections of various specifications, the number of virtual machines of various specifications in the entire resource pool is comprehensively evaluated with reference to virtual machine resource constraints, resource utilization and load balancing. Finally, with the goal of minimizing energy consumption and minimizing the maximum link, the capacity planning in physical servers is implemented under the multi-dimensional resource constraints. Experiments show that the matching model of physical-virtual resources proposed in this paper still has high performance in the case of high requests and processes. This result is conducive to improving the utilization of network resources and reducing business operation costs.
作者
杨敏
徐磊
YANG Min;XU Lei(CETC Potevio Science&Technology Co.,Ltd.,Guangzhou 510000,China;China Unicom Jiangxi Branch,Nanchang 330000,China)
出处
《移动通信》
2022年第12期102-107,共6页
Mobile Communications
关键词
业务预测
虚拟机容量规划
物理服务器容量规划
service prediction
virtual machine capacity planning
physical server capacity planning
