Multi-terminal Control In Cloud Platform

Cloud computing is characterized by resource virtualization,on-demand provisioning,and volume-based charging.SaaS(Software as a Service)is provided to enable users to connect to cloud platforms and use any applications resources at any location and at any time through devices with networking capabilities.This method not only freed itself from geographical restrictions,but also saved user data in the cloud,effectively ensuring the security of user data.These advantages make SaaS services widely used,but with it comes the rapid expansion of cloud virtual machine clusters that provide application virtualization services.For the application deployment and upgrade of cloud virtual machine clusters,the traditional manual deployment mode cannot adapt to the current work.Not only will have a lot of tedious work,but it may also be due to human error in the operation process,leading to the failure of application deployment,reducing the work efficiency of the deployment personnel,and affecting the application on-line.In order to reduce the tedious and redundant work in application deployment,a multi-terminal control tool for Windows Cloud virtual machine clusters under the LAN was designed and developed,to achieve the goal of managing cloud virtual machine clusters on a single host.By providing self-organization of cluster nodes,rapid transmission of application installation packages,and batch execution of commands,it focuses on implementing automatic deployment of applications,reducing human error in the process of deployment and upgrading,and improving work efficiency.This thesis firstly extracts the system requirements according to the specific application scenarios and then introduces the related technologies about design and implementation.For single-node management cloud virtual machine clusters,there are system bottlenecks.The design groups the system nodes and creates proxy nodes for the groups.The agent node is responsible for managing some of the work nodes in the grouping,thereby enhancing the scalability.For the problem of system failure caused by the failure of the proxy,the design uses a distributed election algorithm to determine the agent node.To ensure that the group can generate the agent node in time and continue the original work when the agent node fails.For the problem of slow file transfer from single node to single node,UDP multicast is used for file concurrent transmission,and the TCP supplementary file ensures the integrity of the file,thereby realizing the rapid transmission of files.For the operation process,nodes have unknown errors and cannot be eliminated in time.Remote desktop modules are designed to remotely manage the management node to the target node,providing an effective way to troubleshoot errors.According to the functional requirements,the design defines the basic data structure and message format,extracts the basic network communication interfaces,and divides the system into five major modules: node management,file transmission,command distribution,remote desktop,and system security,to achieve code reuse and module decoupling,and then specifically implement each module to complete the system development.Finally,a cloud virtual machine cluster environment is set up for testing whether each functional module can work normally.At the end,the full paper work is summarized and the subsequent optimization work is listed.