Design And Implementation Of Active Fault Tolerant Service Based On OpenStack Virtual Machine Migration

With the gradual popularization of cloud computing technology,cloud platform reliability is increasingly important.Providing fault-tolerant services for applications running on the cloud platform is a key issue that needs to be addressed.An important feature of the OpenStack cloud platform is the live migration of virtual machines.Virtual machine live migration refers to the virtual machine from the current working physical machine migrates to another physical machine.And,migrating does not affect the normal work of the virtual machine and achieves transparency for users.Through live migration,you can achieve the purpose of fault tolerance.However,the live migration capabilities of the current OpenStack platform can only be done manually by the administrator,with some hysteresis,and when considering the source machine and the target physical machine that are migrated,the considerations are not comprehensive.Aiming at solving the problem that live migration of virtual machines only can be done manually,this thesis presents a fault-tolerant method for active live migration of virtual machines,which is based on the system of physical nodes in the cloud platform.By monitoring and analyzing the resources and hardware of physical nodes,virtual machines on uneven load or failure physical nodes can migrate to the appropriate other physical nodes to achieve active fault-tolerant services.This service can prevent system errors during the process of cloud platform server cluster,where physical nodes in a high load state for a long time.This method can achieve the stability of cloud platform server cluster.The thesis mainly completes the following three aspects:(1)Analyze the overall architecture of the OpenStack cloud platform,study the workflow and source code of the platform live migration function,find the place where the function can be improved,that is,the live migration function in the platform is extended and the virtual machine is active Live migrated.So the design can be based on the OpenStack platform for active live migration of fault-tolerant service system.(2)Introduce the design and implementation of active fault-tolerant service based on OpenStack virtual machine migration.The system is implemented by Python language.It mainly includes four parts:monitoring module,analysis module,communication module and fault-tolerant migration module.The designed processes and the specific implementation methods of the above four modules are introduced.(3)To build the experimental environment and test analysis:build OpenStack platform and set up the platform and NFS installation to ensure live migration function of the correct operation;install Psutil library and Lm-sensors tool on the host to monitor system environment,including system resources and host hardware resources;install the Twisted framework for communication between nodes;install the OpenStack Client to invoke live migration commands of OpenStack platform,and this client mainly provide support for system to schedule virtual machines.After the environment is completed,the module and system are tested.The module test includes testing the four modules of monitoring,analysis,communication and fault-tolerant migration.The system tests include simulating the online shopping system and the real-time broadcast system,to testing overall function.