Incremental backups and differential backups are traditional backup strategies found in backup software but are sometimes also available in storage devices.
It doesn’t matter which software performs it, though there are differences in the interpretation of the terms “incremental” and “differential”.
The “correct” meaning of incremental and differential backup has to do with deduplication.
When Do We Use Incremental Backup?
Incremental backup deduplication is used when you have a large file to back up and it is likely to grow and / or change from day to day (assuming you back up daily).
Incremental backups start off with a full backup, which is usually also compressed.
Monday: Full backup 100GB -> 50GB. Say on Monday you ran the first Hyper-V backup of a huge VM. It was compressed to 50GB.
Tuesday: Incremental backup: 10GB. Due to the changes that occurred between Monday and Tuesday, the backup software found differences and created an increment file, called an in-file delta, which is a file containing the changes between Tuesday and Monday in compressed form.
Wednesday: Incremental backup: 2GB. You can see from Wed to Tue the changes were less in our example. The in-file delta increment is now only 2GB. It refers to Tuesday’s backup.
Incremental backups are very storage efficient because they compare each backup cycle to the previous cycle. Changes are relatively small from cycle to cycle (usually) and the increments are thus small, too.
Downside of using increments: When restoring the backup software needs to process each increment, one after the other. And this is exactly why “differential backup” was “invented”:
When Do We Use Differential Backup?
Differential backup is for those occasions when you need to restore quick and you don’t mind paying a little more for storage in return.
Since there is no method on earth that provides only advantages without adding disadvantages, differential backup has its own, too.
Differential backups restore either in one step (the full backup originally taken) or in two steps (one of the differentials combined with the latest full backup). This helps restoring the backup file much faster but there is a catch: because the differentials are all referring to the last full backup, rather than the last cycle as with incremental backups, the result is (usually) that differential delta files tend to get bigger with each backup cycle as the file changes since the last backup add up. Example:
Monday: Full backup 90GB -> 50GB. Say on Monday you ran the first Hyper-V backup of a huge VM. It was reduced to 50GB.
Tuesday: Differential backup: 10GB. Now the backup software created the first differential, referring to Monday.
Wednesday: Differential backup: 12GB: Now the difference between Wed and Mon grows to 12GB
Thursday: The diff delta file is now 17GB: As you can see the difference is now 17GB because it’s not referring to Wed, it’s still referring to Monday.
The best backup strategy will obviously depend on your needs and infrastructure. And let’s not forget, you could use both strategies simultaneously in two backup tasks.
BackupChain offers both strategies and you could set up two tasks to do both with different retention periods.
When restoring incremental and differential backups, BackupChain automatically puts the pieces together and performs cleanups, too, so you don’t need to worry about full backup drives.
A smart method utilized in BackupChain to cut down on restore time is to create full backups intermittently so the backup chain lengths are limited. When the chain is short, it can be deleted often and cleared from the backup target.