Anatomy of A Backup Chain: Snapshots and the Live Dataset

Topic

This article discusses the Datto backup chain.

Environment

  • Datto SIRIS
  • Datto ALTO
  • Datto NAS

Description

The Datto Backup Solution

The Datto backup solution is a block level backup engine that uses a live dataset and snapshots through dual backup engines.

Advantages of ZFS

  • Copy on Write: When new data is written, previous data is preserved, making snapshots in time possible.
  • Immutability: Snapshots are immutable copies of the file system, meaning once written, the data cannot be changed or edited in any way, regardless of permission levels.
  • Data integrity: Checksum verification and self-repair prevents data corruption or loss.

Live Dataset vs. Snapshots

The live dataset of an agent is the most recent backup image with incremental changes from backup to backup. This represents the most current image that we have for a device. It's essentially the last backup of the agent presented without a snapshot.

After the backup is completed, the live dataset is updated and presented to snapshot. Once that snapshot takes place, you are given a point-in-time backup.

Removing the Live Dataset

There are certain times where it will be necessary to remove the live dataset of a server.

  • When an agent has a particular volume resized, we need to remove the sparse image of the disk to generate a new full image.
  • This accommodates the newly changed volume size. You'll typically see this paired with an error Destination Volume Too Small.
  • A Differential Merge is used during troubleshooting in order to try and repair a backup image.
  • If the image is still not being correctly processed, remove the entire image and process a new one. This will result in a new full image of the volume to be generated and processed.

What removing the Live Dataset won't do

  • Removing a live dataset will not clear up space unless the image has not been completely processed by the device. An example would be when a full backup gets cancelled or interrupted in the middle of the backup attempt. This is uncommon and typically requires tech support intervention.
  • Removing a live dataset will not affect your previous backups.
  • Removing a live dataset will not resolve corruption issues replicating from the production machine, nor will deleting your local backups entirely. If a check of the production environment does not reveal any errors, or if filesystem errors persist in your new backups after resolving production issues, contact Datto Technical Support for further assistance.
  • The backups that you already have on the device will not be affected by this process.

NOTE  For more information on the storage space used by a Datto backup chain, please see Explaining the Inverse Chain and Investigating Large Backups.