An Efficient and Reliable Backup and Recovery Method in Computing Close to Storage Cloud Platform

Abstract

The invention discloses a method for efficient and reliable backup recovery in a calculation approach storage cloud platform. The problems of performance lowering and poor data reliability and system availability caused by data backup recovery in the calculation approach storage cloud platform are solved. The method of combining real-time backup and snapshot and combining local backup and cross-server backup is utilized, the difference snapshot is performed locally, and the snapshot is synchronized to a backup server when primary and standby server resources are idle. For a virtual machine which is high in real-time requirements, backup is performed locally in a file mode through monitoring of virtual machine file changes, and the backup is synchronized to the backup server after a certain time or when primary and standby resources are idle. Thus, not only is running and backup consumption reduced, but also virtual machine and primary and standby server abnormalities can be quickly recovered to a current or specified time point, and the problem of system crash and even complete data loss caused by crash of a distributed file system is solved. Integral performance and high reliability of the cloud platform are greatly improved, and the goal of calculation approach storage is achieved genuinely.

Description

technical field [0001] The invention relates to an efficient and reliable backup and recovery method in a computing close to storage cloud platform. Background technique [0002] With the development of the Internet and mobile Internet, cloud computing technology based on virtualization and other technologies has penetrated into all walks of life. With its characteristics of resource aggregation and on-demand allocation, it is gradually replacing the existing IT architecture. High availability, data security and high reliability are the key indicators to focus on. [0003] In order to ensure high data reliability, high-end SAN and distributed file systems are generally used for data storage, mainly to store virtual machine-related files. The former is expensive, reliable, and poor in dynamic scalability, and is gradually being replaced by the latter. The latter is the now popular Server SAN or vSAN. The main method is to superimpose an open source distributed file system (s...

AI Technical Summary

Problems solved by technology

The reason is that there is a distributed file system between computing and storage. During access to the distributed file system, the client must first find the file location (storage node) through the distributed file system server, and then access the data through the storage node. The data may exist Multiple nodes, these nodes are not necessarily on the same server as the client, if this needs to be accessed through the network, it will undoubtedly increase the delay of IO, and the network IO performance is an order of magnitude worse than that of the local

In addition, in order to ensure high availability, data changes must be backed up to the copy, regardless of the current resource usage, it will be continuously backed up, especially the copy must be written to the hard disk of another server through the network, and the next IO operation can only be performed after receiving a response. The intermediate delay will be relatively large, especially when encountering congestion, which will greatly affect the response speed of the virtual machine, which is fatal to applications such as virtual desktops that occupy high IO resources and require high user experience.

[0004] In addition, the reliability of the distributed file system is only relative. Once the distributed file system crashes and cannot be recovered, all server data will be lost, which will be a catastrophic consequence; and virtual machine data loss (such as accidental deletion, attack) and other The copy is also lost in real time, and the copy is unreliable

Images