摘要
面对算力资源快速交付需求的持续攀升,云资源提供方通常面临诸多挑战。一方面,用户需要能够即时获得具备弹性差异化配置的容器资源;另一方面,还需保障算力交付链路具备高可用性、强稳定性以及高容错性。为此,提出一种利用资源池技术的批量容器算力交付框架,探究容器在不同工作负载之间实施迁移的可行性,并清晰阐明基于可配置控制器组件的容器生产链路设计方案。实验结果表明,所提框架契合设计原理,并验证了借助资源池技术的容器生产链路具备实施快速资源交付的能力。
In response to the ever-increasing demand for the rapid delivery of computing power resources,the cloud service providers are usually confronted with multiple challenges.On one hand,users need to instantly obtain container resources with the elastic customized and differentiated configurations.On the other hand,it is also necessary to ensure the computing power delivery chain is equipped with high availability,strong stability and high fault tolerance.To this end,a batch container computing power delivery framework utilizing resource pooling technology is proposed.It explores the feasibility of migration of containers between different workloads,clarifies the design scheme of the container production chain based on configurable controller components.The experimental results show that the proposed framework conforms to the design principles.Additionally,the experiments confirm that the container production chain leveraging resource pooling technology is capable of rapid resource delivery.
作者
施凯
丁宇
范贵生
SHI Kai;DING Yu;FAN Guisheng(Department of Elastic Computing and Container Service Team,AliCloud,Alibaba Group,Shanghai 200030,China;Department of Cloud-native Application Platform,AliCloud,Alibaba Group,Hangzhou 310030,China;Department of Computer Science and Engineering,East China University of Science and Technology,Shanghai 200237,China)
出处
《微型电脑应用》
2025年第6期199-203,共5页
Microcomputer Applications
关键词
容器算力
容器迁移
资源池
Kubernetes
弹性计算
container computing power
container migration
resource pooling
Kubernetes
elastic computing
