A method of enabling File-level restore from Image-level backup
Publication Date: 2010-Aug-30
The IP.com Prior Art Database
Disclosed is a mechanism for enabling virtual block device mapping by combining the meta data from catalog and the storage media from data repository, which allow a backup system to provide individual file-level restore from image-level backup. Regarding a backup system, users require their file system to be backed up quickly (in a smaller backup window), and individual files can also be restored quickly in case of the file is damaged or deleted by mistake. There’s a tradeoff between backup speed and restore speed for a traditional backup system. 1) Use Image-level backup/restore to fulfill the requirement of backup speed. I. The whole file system (block device) is backed with high backup speed, into a single backup image by blocks. II. Have to restore the whole file system from image backup in order to restore a file, which results in very long restore time. Also leads to the waste of network bandwidth and other resources. 2) Use File-level backup/restore to fulfill the requirement of restore speed. I. Files can be restored individually very quickly. II. Have to backup all the files one by one, which results in much slower backup speed. Also leads to much more I/O and computing resource will be involved for clients with huge number of small files (e.g. disks used by exchange server). It’s ideal to achieve both advantages from these two backup/restore mechanisms - high backup speed of Image-level backup and high restore speed of File-level restore. But in a traditional backup system, they don’t work together. This article will introduce a new method, which will allow individual file restore from Image-level backup, to fulfill the requirement of backup speed and restore speed.
A method of enabling File -
files need to be read and transferred during the restore.
In conclusion, this new method enables doing file-level restore from image-level backup of the traditional backup system, which means to achieve both high backup speed and restore speed at the same time.
Here's a snapshot of how the proposed VDM works:
The VDM simulates the virtual disk image based on the below data:
1) The backup catalog (database), that records which image has been backed up and stored in which backup media. The catalog is created and maintained by backup system.
2) Backup media, that stores the data backed from the client. Various kinds of backup media can be used for backup, including file and tape.
A disk image is essentially a block device. During an Image-level backup, the backup system copies thedata from the client disks to the backup media, and builds the catalog to describe the relationship between the client disk and the backup media.
Table.1 is an example the catalog of a full image backup of Disk1. In this example, Disk1 contains 1000 blocks, and was copied onto 3 segments on 2 tape volumes.
-level restore from Image
level restore from Image -
This disclosure proposed a mechanism, Virtual Disk Mapper (VDM), which can be used to simulate a virtual disk image (RAW block device) based on the catalog and media created during backup. This virtual disk image can be mounted locally on server or exported to client, and client can simply mount the exported disk image with appropriate file system format and restore only the files they need.
With this proposed method, users can:
1) Apply image backup to ensure high backup speed. The Image backup function exists in most popular backup systems.
2) Used VDM described above to get a virtual disk image exported from backup system, and restore the files at will.
Note that only the selected
Disk1 0 500 Tape01 500 Disk1 500 400 Tape01 1000
Disk1 900 100 Tape02 200
The Virtual Disk Mapper (VDM) maps the data on tape volumes or other backup media to a virtual disk.
A file read/write request against...