Font Size: a A A

Design And Implementation Of A Cluster Management System Coordinated By Private Cloud And Public Cloud

Posted on:2021-03-01Degree:MasterType:Thesis
Country:ChinaCandidate:Y D F OuFull Text:PDF
GTID:2428330623467797Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the rapid development of cloud computing,the functions of cloud service products developed by major Cloud Service Providers have improved gradually.Cloud service providers provide users with abundant computing resource pools and storage resource pools,which makes users are able to purchase on demand.Public cloud services take advantages of conveninently applying,allocating relying on demands and supporting autoscaling.Based on the advantages of conveninently opening,on-demand distribution and elastically scalable resources in public cloud services,more and more traditional companies deploy their applications with large changes in the elasticity of resource requirements to the public cloud servers.As for applications with stable resource requirements,the companies prefer to delpoy them to the private cloud servers and utilize computing resources of pulibc cloud to provide more stable IT services.Recently,containerization of IT services has become increasingly common.Docker container technology has the characteristics of quick start and portability.Thanks to the docker technology,IT developers can freely deploy acclications,and use container cluster management tools to schedule resources and orchestrate management on multiple container services.According to the load of computing nodes,container services can be flexibly mapped to appropriate computing nodes to improve resource utilization and enhance the ability of IT services to respond to changes in demand.Therefore,based on container cluster management technology,a cluster management system that coordinates private and public cloud computing resources has wide application value.This thesis uses the cloud platform of the laboratory as the research object,and conducts comprehensive analysis of the cloud platform architecture and functions and finding many aspects that can be optimized,such as virtualization service cluster management,universe storage management,cloud disaster recovery,and transferring from private cloud cluster to public cloud clusters seamlessly.This thesis has thoroughly researched on the existing hybrid cloud architecture solutions along with their core technologies,Kubernetes cluster management technology and Ceph distributed storage technology.After understanding technic details,this thesis proposes a collaborative cluster management system combining private and public cloud services,called privatepublic collaborative cluster management system.In this thesis,the central control management layer and the daemons are designed to form a master slave architecture to build a host cluster.The host cluster treats servers whether from private or public cloud as universe computing nodes and manages them via the central controlling layer.The central controlling layer encapsulates public cloud APIs in order to autoscale dynamically resources of the public cloud.Data center is deployed distributedly on both private and public clouds for the storage of image files,configuration environments and application records.Ceph storage cluster is adopted as the back-end of data center.The synchronization solution of Ceph helps to migrate data from the data center of private clouds to that of public clouds and to keep synchronization.In this way,the design of data center implements the feature of cloud disaster recovery mechanism.For the host cluster,multiple Kubernetes clusters are constructed according to the demand of diverse cloud environments.The central controlling layer interacts with Kubernetes clusters through a daemon process on the computing node,so that the central controlling layer can jointly manage Kubernetes clusters in private and public clouds.With evaluation,the cluster management system designed and implemented in this thesis can effectively manage the computing resources of private cloud and public cloud,rapidly deploy cloud services,schedule private cloud computer nodes and public cloud computing nodes to work together,and provide reliable cloud services.
Keywords/Search Tags:private cloud, public cloud, resource scheduling, cluster management
PDF Full Text Request
Related items