基于动态规划的降低Openstack资源碎片率的方法  

A Method for Reducing OpenStack Resource Fragmentation Based on Dynamic Programming

作　　者：杨松[1] 

机构地区：[1]中兴通讯股份有限公司无线与算力研究院,江苏南京210012

出　　处：《工业控制计算机》2025年第3期105-106,128,共3页Industrial Control Computer

摘　　要：在虚拟化领域,为降低云平台资源碎片率,虚拟机一般采用集中部署的方式,通过穷举法预部署,为批量虚拟机预先规划好每台的部署节点,之后按照预部署结果执行部署以减少碎片率。但在实际应用中,虚拟机规格特征参数较多,随着虚拟机数量增加,参与穷举的场景呈指数级增长,导致工程应用性较弱。为解决这一问题,通过提取影响碎片率的主要特征参数作为虚拟机单一规格特征,并采用动态规划的方式对虚拟机进行预部署确定部署节点,同时通过与云平台之间增加确认机制保障预部署节点的正确性,从而降低云平台资源碎片率。In the field of virtualization,we concentrate on deploying virtual machines in batches to reduce resource fragmentation.Through exhaustive pre-deployment,each deployment node for batch virtual machines is pre-planned,and then deployment is carried out according to the pre-deployment results to reduce fragmentation.However,in practical applications,virtual machines have many characteristic parameters,and as the number of virtual machines increases,the scenarios involved in exhaustive enumeration grow exponentially,leading to weak engineering applicability.To address this issue,this paper extracts the main characteristic parameters affecting fragmentation rate as the singular characteristic of virtual machine specifications,and uses dynamic programming to pre-deploy virtual machines to determine deployment nodes.Additionally,a confirmation mechanism is added between the virtual machines and the cloud platform to ensure the correctness of the pre-deployment nodes,thereby reducing the resource fragmentation of the cloud platform.

关 键 词：虚拟化 预部署 动态规划 资源碎片率 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]