Method and device for copying image file

Abstract

The embodiment of the invention discloses a method and a device for copying an image file. The method comprises the following steps of when the image file of a production center is copied to a data backup center, obtaining the target position information of a copied data block of the current image file; when a write request on the image file is received, and the number corresponding to the starting position of the write request is greater than the number corresponding to the target position, directly executing the write request; when the number corresponding to the starting position of the write request is smaller than the number corresponding to the target position, executing the write request, and recording the write request into a pre-established write request log; when the copying of the image file of the production center to the data backup center is completed, transmitting the data of the write request log to the data backup center, and replaying, so as to form an effective image snapshot. The method disclosed by the embodiment has the advantages that the influence to the I/O (input/output) request operation of an upper layer is low in the image file copying process, the log copying operation is reduced by one half, and the copying efficiency of the image file is improved.

Description

technical field [0001] The invention relates to the technical field of storage disaster recovery, in particular to a method and device for mirror image file replication. Background technique [0002] The hyper-converged all-in-one machine is an IT infrastructure component that integrates virtualization management software, virtual storage, virtual network and other basic components, and uses standard X86 server hardware to carry virtualization management software, virtual storage, virtual network and other software. Hyper-converged all-in-one machine storage adopts a distributed storage architecture. Before use, a cluster needs to be established. The physical disks of all all-in-one machines in the cluster are virtualized into storage resource pools, and then storage services are provided to virtualization management software. Through the virtualization management software, the virtual machine can share the same physical computer, but for the virtual machine, it runs in the ...

AI Technical Summary

Problems solved by technology

[0005] There are problems and deficiencies in the existing image file replication technologies based on snapshot technology. For example, the copy-on-write technology affects the write operation of the upper-layer business, or has a great impact on write-intensive business. The write operation adds a write operation, which increases the write operation delay of the system, and also brings disk random access overhead for mirror file copying; while I / O redirection has a great impact on read-intensive businesses, because The write operation request is written to another piece of storage space, so each read request needs to find out whether its data has been redirected, and the originally possible sequential read request operation has become a disk random access, which increases the response time of the upper-layer business of the system. In addition , after the mirror file is copied, the I / O redirected data needs to be merged additionally, which also increases the impact on the upper-layer business

Images