PlanetScale introduces Rewind feature to ‘undo’ bad schema migrations

PlanetScale is adding a Rewind feature to its serverless MySQL database-as-a-service, allowing users to ‘revert changes’ up to 30 minutes after a schema migration deployment request is made.

PlanetScale wants to help database administrators or devops professionals to easily ‘undo’ bad schema migrations earlier, without the pain of manually rolling back and causing downtime or data loss issues for the business.

“PlanetScale Rewind gives users the ability to undo a bad migration just as easily as you undo a typo,” said Nick Van Wiggeren, vice president of engineering at PlanetScale. “Instead of having to go restart the entire process, imagine if you could just click a button — like Control-Z — and get it back to where it was when you started but, notably, without losing a byte of data.”

PlanetScale

While other database-as-a-service providers have given users the ability to revert to a snapshot, which typically entailed losing some data during the deployment or migration process, PlanetScale claims its Rewind feature is an industry first due to the speed at which customers can reverse changes without losing data.

The feature is powered by VReplication, a native capability of the open source clustering and management system Vitess, which underpins the PlanetScale database and which enables faster migrations and therefore quicker rollbacks as needed.

“VReplication can be setup to migrate data from an existing system into Vitess. The replication could also be reversed after a cutover giving you the option to rollback a migration cutover if something went wrong, without losing the writes to the migration target,” a Vitess blog post claims.