Point in Time Restore (PITR)

MemSQL supports incremental backup, but not Point in Time Restore (PITR). If, for example, you made a user error that requires restoring the database, depending on how often your incremental backups are run to capture the deltas, you may run into a situation where you lose some data between the last incremental backup and the time that the error occurred.

This feature will enable a user to recover their database from a point-in-time and will decrease the recovery point/time objective. Additionally, it is present in other database systems.

Note: A member of the MemSQL team has filed this feature request. We are adding in several features to this forum that we believe are popular with our users to engage and better understand the needs of the MemSQL community.

PITR shipped as a SingleStoreDB feature in 2021. It requires use of unlimited storage databases (bottomless):


1 Like

It’s nice to have the ability to recover from user mistakes by providing PITR with Unlimited Storage Database.

However, the PITR function using database backup is still required. Where I live, companies in the financial sector must backup their databases to media and store them in an independent space. For this reason, full/incremental backup operations are very customer friendly operations. In a situation where customers are familiar with PITR using backup, it is not easy to accommodate system architecture changes for PITR. In addition, financial companies have many restrictions on the use of public cloud, making it difficult to use public cloud object storage.

For other reasons, it is not easy to immediately apply PITR using object storage in other sectors such as manufacturing.
Many customers who cannot apply Unlimited Storage Database still need the PITR function with backup.