Design And Implementation Of The High Availability For Desktop Cloud Platform

It has been more than a decade since the concept of cloud was proposed.During this period of time,Cloud computing experiences from the initial concept to the current combination with all walks of life.Governments and enterprises have seen the evolution of IT system that the cloud computing will bring about.Cloud computing,with features of unified control,energy efficient,safe,flexible telescopic and low-cost,is now widely accepted by governments and enterprises.They urge to seek cloud,which makes their IT system became more lightweight,flexible,controllable,and in the meanwhile,reduces the cost.Also,the desktop cloud,as one of the cloud service model,has won the love of the enterprise.But all things are both good and bad.In the large-scale use of the desktop cloud,the desktop cloud itself has its own problems because of its dependence on the network.Downtime of Cloud service occurs frequently,which hinders the spread of the desktop cloud.In this context,this thesis seeks to improve the overall availability of the desktop cloud system by improving the architecture and implementation of desktop cloud.On the basis of analyzing a large number of desktop cloud products,this thesis mainly does the following work.1.Researching the overall implementation of the desktop cloud architecture and designing a new high availability desktop cloud system architecture.The improvement is mainly reflected in the high availability of computing virtualization and storage virtualization.2.Introducing the multi-user to multi virtual machine mechanism.Virtual machine will work as virtual machine pool,which improves availability and resource utilization.3.Introducing new state decision rules of virtual machine and virtual machine pool,which ensure high availability of scheduled virtual machine and that the new connection of users do not affect the original user's experience.4.In view of the irrationality of default allocation of resources,allocated resources will be adjusted in real time according to the specific situation,which improves service level and resource utilization.5.We design a new distributed file system architecture and abandon the original backup design of name-node,so that the name-node can be linearly expanded,which improves the high availability and concurrent of storage system.