Traditionally, there have been three types of backups: full, incremental and differential. All of these types have their strength and weaknesses and therefore, administrators have to weigh each option to see which works best for their scenario. While these three methods are most common and widely used, this article introduces a fourth type: reverse incremental.
An important feature when backing up important data is the ability to restore previous versions. Therefore, every type is compared with respect to its ability to maintain several versions of the data. .
As the name implies, a full backup is an exact replica of the source files.
|Easy Restore - Restoring files is very easy since there is only one version available.||No Automatic Versions - You can only restore the most recent copy|
|Extra Storage - If versions are needed, full copy of the backup must be stored on different storage location, hence wasting disk space.|
Incremental backup retains a full copy of the initial backup and subsequent backups contain changes since the last backup. For example:
|Multiple Versions - Versions are created automatically||Potential Data Loss - Since the most recent version of the backup relies on a previous backups, a file corruption can render the backup useless. For example, in order to restore Thursday's backup, you will need to first merge Monday with Tuesday, then with Wednesday and finally with Thursday to construct the full backup on Thursday.|
|Saves Disk - Since subsequent backups will only retain changes, the actual delta between two backups is relatively small.||Limited Versions - In order to avoid data loss, a full backup must be performed at regular intervals.|
|Longer Time/Bandwidth - Since entire backup gets transferred, it not only takes a long time but also consumes significant network resources if the destination is not on the same machine|
Differential backup are similar to Incremental with a subtle difference: subsequent backups do not rely on intermediate copies. Instead, they are built on top of the first full backup. For example:
|Multiple Versions - Versions are created automatically||Potential Data Loss - Although there is still a chance for data loss, the probability is reduced as compared to incremental.|
|Saves Disk - Similar to incremental, this method also saves disk space but not as much as incremental.||Limited Versions - Full backup must be performed at regular intervals. Failure to perform full backup will result in large deltas, which will eventually defeat the purpose of having differentials.|
This is similar to incremental backup but in reverse direction. Meaning the most recent copy of the backup is always saved as full. Previous versions are stored as deltas on the destination. For example:
|Multiple Versions - Versions are created automatically||There are no disadvantages in using this type of backup.|
|Saves disk - Significant storage savings since only the most recent version is maintained as full. Older versions are just the changes.|
|Unlimited Versions - Since the versions go backwards, there is no need to create a full backup periodically. Therefore, you can have unlimited versions.|
|Faster Restore - In most cases, users want to restore the latest version, which is stored in its entirety on the destination. Therefore, accessing backed file on the destination is very easy.|
|Data Corruption for versions||N/A||3||3||4|
The chart displayed above clearly shows the Reverse Incremental method is the best. One way to create such backups is to use Hard Links on Linux/Unix. However, using this technique is not for everyone as it requires in-depth knowledge of how the file system works on the OS.
|Created on:||Oct 9, 2020|
|Last updated on:||Dec 4, 2023|