There are three types of backup strategies, full, incremental and differential. A full backup backs up everything, regardless if it has been backed up before or not. A restore from a full backup is easy because everything is available and there are no dependencies on other backups. The trade-off for this ease is the time it takes to perform the backup. Since everything is being backed up, this type of backup takes the longest period of time.
An incremental backup only backs up things that have changed since the last backup, either full or incremental. A restore from an incremental backup requires the last full backup and every backup since the last full backup. This means a restore from an incremental backup chain (what the series of backups since the last full backup is called) takes longer than a restore from a full backup. However, the backups themselves are much shorter than a full backup since it is assumed that not everything has changed since the last backup, full or incremental.
A differential backup backs up all data that has changed since the last full backup. A restore from a differential backup strategy only requires the last full backup and the last differential backup. The time it takes to do a differential backup increases as the time from the last full backup has occurred. This is because it is assumed that the amount of data that has changed since the last full backup increases as time goes on.
In my last post I talked about the types of backups. In this post, I’ll talk about backup strategies and scheduling. Whether you are doing an image backup or a file backup, you have to decide on a backup strategy and the backup schedule that implements this strategy.
Backup strategies don’t care what type of backup you choose. A backup strategy determines how you are going to handle data that has changed since the last successful backup. There are three basic backup strategies – full, incremental and differential.
A backup in a full backup strategy copies everything whether it has changed or not. This strategy results in the longest backup time for each backup made, since everything defined by the backup type is copied.
A restore from a full backup only requires the information from the backup and nothing else.
A backup in an incremental backup strategy requires a full backup. Then each subsequent backup copies anything that has changed since the last successful backup. Since each backup only contains the latest changes from the previous backup, this strategy completes the fastest for each backup between full backups.
A restore from an incremental backup strategy requires the information from the last full backup and every other backup until the backup you want to restore from. For example, let’s assume you do a full backup on Sunday and incremental backups daily. If you want to restore data from the Thursday backup, you would need data from the Sunday, Monday, Tuesday, Wednesday and Thursday backups to complete the restoration. It would be a safe assumption to say that an incremental backup strategy has the longest restoration time.
A differential backup strategy also requires a full backup, but in this case, each backup copies anything that has changed since the last full backup. Since all the changes from the last full backup are copied in this strategy, each backup between full backups takes longer and is bigger than the backup before it. This is because it is assumed that the number of changes in the data being backed up increases as time goes on.
A restore from a differential backup would require information from the last full backup and the last differential backup prior to the desired time of the data instance to be restored. So, using the prior example of a restoration from the Thursday backup, you would only need data from the Sunday and Thursday backup.
The reason for the multiple strategies is really dependent on the backup package used and the time it takes to do each of the backups. Full backups take the longest. Incremental backups are typically faster (because there is less data to copy) than differential backups. Generally, the time to perform a restore is not considered since it should not be required on a frequent basis.
Backup schedules really depend on the business and the time it takes to do the backup. When I started in this business many years ago and backups took all night or more, the gold standard for backups was to perform daily backups that were kept for four weeks. Weekly backups (think the full backups) were kept for 2 months, monthly backups (first or last weekly of the month) were kept for a year. A similar schedule is still considered a gold standard.
In these days with ransomware being one of the leading types of malware attacks, it is recommended that periodic copies of backups be kept offline.