Design And Implementation Of A Two Level Resource Scheduler Based On Kubernetes-on-EGO

Cloud computing is a new technology,it makes the physical hardware,network resources abstract as a huge pool of resources through the virtualization technology,deal with the enterprise data center,makes the basic service facilities,a variety of software and platform resources into services In the form of the network available to the user.Now large-scale cluster system is emerging,how to manage large-scale data center resources is a problem,in this case produced cloud platform management software.With the widespread use of container technology,Kubernetes,a new generation of container cloud-based solutions,came into being.Kubernetes is a new container-based distributed computing platform that enables Kubernetes to automate resource management to maximize resource utilization across data centers.Resource scheduling is one of the core modules of this large-scale data center,a good scheduling system can select the available resources for the user according to the user's request,and can improve the utilization of resources,and Kubernetes' resource scheduling system is still There is a lack of large-scale data center can not meet the needs of users to configure their own resources,resource scheduling inefficient,can not be used in the actual production environment Kubernetes cloud management platform.Therefore,based on Kubernetes-on-EGO,this paper designs and implements a resource scheduler with two-level scheduling architecture for Kubernetes.In this paper,Kubernetes and the EGO container cloud resource scheduler are deeply researched and found that Kubernetes is inefficient and functional in resource scheduling.The user can not configure the resource scheduling scheme.The algorithm is too simple to design.The architecture is a simple single scheduler architecture EGO has a rich scheduling algorithm,belonging to the two-level scheduling architecture,the user can configure the file on the scheduling plan,scheduling task priority configuration.Based on the in-depth study of the two schedulers,this paper designs and implements a resource scheduler based on Kubernetes-on-EGO two-level scheduling architecture.First,we study Kubernetes' overall architecture design,the principle of dispatching system and the realization principle of EGO scheduling system.How to integrate the two platforms and develop the dispatch system of Kubernetes and EGO.The key to integrating two scheduling systems is to establish dynamic shared library,EGO scheduling system package API,the establishment of EGO GO language based on the API,in the Kubernetes scheduling system called EGO based on the This paper studies the design mechanism,system basic concept and service design principle of Kubernetes cloud platform,and introduces the technology of cloud computing platform resource scheduling,and then analyzes the business process of dispatching system and designs the module.By drawing the class diagram,The design of the integrated dispatching system is designed,the system is divided into different functional modules,and the modules are elaborated by the class diagram.Finally,the design and implementation of the cloud computing platform Kubernetes scheduling system is completed based on the EGO scheduling system.Based on the Kubernetes and EGO integrated platform test results show that the improved scheduling system platform greatly improves the efficiency of resource scheduling and resource utilization,and users can flexibly according to their needs on the scheduling system configuration,to achieve the expected aim of design.