File-based versus Image-based Backups
This document provides a comprehensive comparison of file-based and imaged-based backups with these considerations in mind:
Storage space efficiency
File-based solutions basically perform backup file by file. In order to support disaster recovery, file-based backup solutions like Retrospect do more than that, as explained later in this document. Image-based solutions primarily back up used blocks on the disk.
Not all file-based solutions offer the same level of protection. To greatly simplify disaster recovery, Retrospect protects different types of data:
User data files
System and application state
Disk layout and boot information for Windows systems
Image-based solutions can be technically simpler because they back up used blocks, capturing the exact state of the disk. File-based solutions are more complicated but can support disaster recovery by backing up every file and folder, including their access rights and other attributes.
Retrospect supports disaster recovery.
Storage space efficiency
File-based backups often require less storage due to a number of reasons.
Excluding unneeded files – OS and applications routinely create various temporary, cache and log files that are unnecessary for disaster recovery. On Windows, some of the large files created by Volume Shadow Copy Service (VSS) and other Windows features under the System Volume Information folder are unnecessary for disaster recovery.
Several image-based products can skip disk blocks occupied by the OS’s page file and hibernation file, but not blocks of many of the other large unnecessary files.
Not affected by defragmentation – Mac OS and Windows automatically move files to reduce file system fragmentation.
This causes disk block changes which are backed up unnecessarily by image-based products.
Flexibility on what to back up – File-based backup provides great control for selecting what not to back up, further reducing storage space requirement.
Image-based backup lacks this flexibility and is larger.
Retrospect backups are much smaller than image backups.
Image-based products access used blocks sequentially, making full system backup and related restore very fast. In contrast, file-based solutions may need to seek back and forth to access each file, thereby lowering performance. This disadvantage may be partially mitigated by OS’s automatic defragmentation, file system level caching and solid state device. For incremental backup and restore, it is hard to say whether image-based approach or file-based would take less time, because performance heavily depends on specific product implementation and incremental backup format. Image-based solutions have the advantage of higher disk IO throughput while file-based solutions have the advantage of processing less data.
For first backup, image-based product is usually faster. For incremental backups, speed depends on specific product implementation.
Backing up VM’s virtual disks (e.g.
vhd) is effectively image-level backups. For information on the benefits of file-level backups for VMs, see our article.
Last Update: 09 septembre 2014