Font Size: a A A

Design And Implementation Of Application-oriented Elastic Scaling Method For Container

Posted on:2021-03-13Degree:MasterType:Thesis
Country:ChinaCandidate:P R ShanFull Text:PDF
GTID:2428330602997217Subject:Software engineering
Abstract/Summary:PDF Full Text Request
With the new round of technological change in the field of cloud infrastructure and the practice of domestic and foreign major manufacturers,container technology has become the mainstream technology of the new resource virtualization computing model.As a popular computing engine technology,Docker has risen rapidly in recent years,effectively solving the problems of traditional virtual machine technology,such as low resource utilization rate,and inconsistent stack of local and remote environment.In the cloud environment,large-scale container cluster orchestration tool is needed,and the corresponding large-scale distributed container cluster management scheme arises at the right moment.Kubernetes technology has become the DE facto standard for the construction of container cloud platform in the industry due to the advanced concept of distributed container orchestration and the secondary innovation ability fostered by the open source community.This paper makes an in-depth analysis of application choreography scenarios and business requirements,Docker container technology and Kubernetes cluster choreography technology architecture,principle and scaling strategy,etc.The automatic scaling method based on kubernetes framework is designed and implemented.Meanwhile,aiming at the problem of insufficient support in the existing methods in the industry on the "custom" index of triggering scaling and the combined use of different schemes,the design scheme of dynamic scaling method of container cluster,namely the design of elastic scaling system,is proposed.In this system,to solve the problem of "business" index missing in the index needed for scaling,a custom index elastic scaling scheme is proposed.On this basis,a "monitoring module" is designed to complete the development of integrated schemes such as business level index definition and acquisition,event monitoring and real-time display.Furthermore,the horizontal expansion module,cluster expansion module and other schemes are used together to extend the elastic scaling from the "horizontal" direction to multiple dimensions.Through the collocation of the above two schemes,the optimized design of "scaling module" is completed.On this basis,combined with dynamic storage,scalable response,visualization,alarm and other support modules,as well as index fusion,scheduling strategy,and so on,thus realizing the overall innovative scheme of the whole elastic scaling system.Finally,a kubernetes based distributed cluster is developed by using multi-node distributed system technology,and the design scheme is tested and verified.The experimental results show that the elastic scaling design scheme in this paper can run stably in the distributed platform of private cloud and meet the elastic scaling requirements under different scenarios.At the same time,it enhances the high availability of application cluster,makes resources and cluster components full of flexibility,and optimizes the utilization mode and efficiency of cluster resources.
Keywords/Search Tags:Docker, Kubernetes, Elastic scaling, Cluster monitoring
PDF Full Text Request
Related items