Search

Retention

Retention is the concept of determining what data should be kept, for how long, as well as when and what data can be safely removed in order to storage space. Retention rules can be configured for individual Protected Items, or for Storage Vaults.

Protected Item Retention

In the Retention section of the Protected Item, you can configure a retention policy to apply when backing up this Protected Item to a specific Storage Vault. If no policy is configured for a specific Storage Vault, the default retention policy for the Storage Vault (30-days) will apply.

The retention section will display (default) to indicate that the Storage Vault default rules apply, Keep (X rules) to indicate that specific Protected Item retention rules have been applied.

Retention (Storage Vault)

Storage Vault retention rules are the default for all data stored, unless a Protected Item has its own set of retention rules. When a Storage Vault is first created, the default, no-change-by-policy retention rule is 'Keep all data forever'. This can be automatically altered by user-profile policy, or by later manual changes.

If a Protected Item is given its own set of retention rules, these Protected Item rules will normally take precedence over the Storage Vault retention rules.

If a Protected Item is removed, or has its retention rules removed, then the Storage Vault retention rules will take precedence once more, and the data associated with the Protected Item will be kept, or removed, in accordance with the Storage Vault retention rules.

Example usages of retention rules:

  • Change the default Storage Vault retention rule to 'Keep all data for 60 days', plus
  • add a Protected Item rule to keep all snapshots for 90-days, plus
  • add a Protected Item rule to keep a representative snapshot from each week, on a Monday at 6am, for 2 years.
  • A customer permanently revokes a single device. The device is revoked from the user-profile list of devices. Once the device is deleted, the Protected Items for the device are also deleted, as well as any retention rules. The snapshot data will now be governed by the Storage Vault retention rules. If no changes have been made, the default Storage Vault retention rules are 'Keep all data for 30-days', so the unwanted data from the Protected Item would be completely deleted by the 31st day.

Explanation

  • When you revoke a device, it will remove all its Protected Items, and all of their retention rules.
  • If the user-profile still has at least one live device that stores data in the Storage Vault, it will run the retention pass into the Vault, eventually deleting the old Protected Item data.

Example Solution

  • Set all Storage Vaults to have a changed-default retention rule of 'keep all data for 30 days', or some other period.
  • When a device is revoked, or a Protected Item is removed from the list of Items to protect, the next retention pass will fallback to the Storage Vault rules.
  • A Storage Vault retention period of 30 days will allow for mistakes to be discovered. A mistakenly-deleted Protected Item can be reinstated into the list of things to protect; or the snapshots of the Protected Item can be restored via another device registered to the same user-profile.
  • A Storage Vault retention period of 30 days will allow for genuinely-unwanted data to be automatically removed at the end of 30 days, thereby keeping storage sizes to a minimum.

Retention Pass

A "retention pass" is the act of cleaning up data from the Storage Vault that exceeds the configured retention policy.

During a retention pass, the desktop application looks at each backed-up job within the Storage Vault and determines whether it meets the retention policy. If the retention policy states that the backed-up job can be safely removed, the backed-up job is removed from the Storage Vault. Once all backed-up jobs are checked against the retention policy, any data chunks that are no longer referenced by a backed-up job can then be pruned to save disk space.

Automatic retention passes

A retention pass may run automatically after each backup job.

An automatic retention pass is not necessarily critical, and so if a backup job comes to an end without a retention pass being possible to run, the retention pass is not immediately "overdue".

The "overdue" rules for automatic retention passes are based on the time since the previous successful retention pass, and, the number of jobs exceeding the configured retention policy. The exact rules are still being determined and are subject to change in future versions.

As of version 21.9.3, the rules are as follows:

If the device is a "high power" device:

Last retention pass 0-2 jobs exceeding policy 3-9 jobs 10-49 jobs 50+ jobs
less than 24 hours ago None Attempt Attempt Require
between 24 hours ago and 14 days ago None Attempt Require Require
between 14 days ago and 21 days ago Attempt Attempt Require Require
over 21 days ago, or never ran Require Require Require Require

If the device is a "low power" device:

Last retention pass 0-2 jobs exceeding policy 3-9 jobs 10-49 jobs 50+ jobs
less than 24 hours ago None None None Attempt
between 24 hours ago and 14 days ago None None Attempt Require
between 14 days ago and 21 days ago None Attempt Attempt Require
over 21 days ago, or never ran Attempt Require Require Require

In the above tables,

  • "None" means that no automatic retention pass will be attempted at the end of the backup job
  • "Attempt" means that eazyBackup will attempt a retention pass, but not throw an error if it could not be performed
  • "Require" means that eazyBackup will attempt a retention pass, and will throw an error if it could not be performed

A device is considered a "high power" device if it meets 2/3 of the criteria:

  • over 7 days uptime
  • running Windows Server or Linux
  • having over 8 GB physical RAM

Manual retention passes

You can run a retention pass for a Storage Vault on demand, by right-clicking the Storage Vault within the backup application.

In this case because the action was explicitly taken, an error will be raised if the job fails to be performed for any reason.

en_USEN