Virtual machine on-line transfer method across virtualization data centers

Abstract

The invention provides a virtual machine on-line transfer method across virtualization data centers. The method includes that step 1, a plurality of virtualization data centers are deployed on a wide area network (WAN), and each virtual machine mirror file is divided into a template and user private data files, step 2, the virtualization data centers are connected through a virtual private network (VPN) of a two-tier bridge pattern, step 3, dirty blocks written into a disk every time are monitored in a virtual machine input / output (I / O) drive course on a source node, a transfer thread is created for the drive course, and the transfer thread pre-transfers the user private data files and transfers monitoring records of the dirty blocks, step 4, a transfer course of a virtual machine on the source node controls the pre-transfer and the transfer of the monitoring record, and step 5, the execution sequence of a destination node recovery course of the virtual machine is changed, a recovery thread is created for the recovery course, and the recovery thread receives information about the pre-transfer and the transfer of the monitoring records and writes the information into a storage node. The virtual machine on-line transfer method across the virtualization data centers can achieve dynamic mapping of virtual machines and resources among the plurality of virtualization data centers.

Description

Technical field [0001] The invention relates to the field of resource management related to virtualized data centers, in particular to a method for online migration of virtual machines across virtualized data centers. Background technique [0002] The owners of traditional IT resources provide services to the outside world in the form of data centers, which are composed of multiple data centers with a large number of computing or storage resources distributed in various places, and different users share the resource pool of the data center. In the cloud computing model, users’ applications are complex and diverse, and their needs are generated at different locations. Different users have different requirements for the quality of service of the application. Therefore, IT companies face the difficulty of managing multiple data centers across regions at the same time. Therefore, the traditional data center management model lacks a method for dynamic adjustment of resources among mul...

AI Technical Summary

Problems solved by technology

However, there are the following problems: when a virtual machine is migrated to a service node that is far away from the virtualized data center where the file system is located (compared to the local file system), the average I / O access latency and maximum access latency of the virtual machine increase. It is more than twice the original size, and the CPU utilization rate is also reduced by 30%. After the virtual machine is migrated, it cannot fully restore the original execution efficiency.

[0008] 2) Virtual machine migration in a cross-virtualized data center environment must send all status information of the virtual machine (including: CPU, memory, disk, etc.). In order to support the different needs of various users, different disk migrations are used in the prior art Strategy, the disk space of the virtual machine is large (more than 10G), and the dirty sectors generated during the state consistency process of the additional disk and the dirty page information generated by memory iteration make the amount of data sent during the entire migration process too large and consume too much network Bandwidth resources, and the virtual machine migration time and downtime are too long

Images