Databricks has introduced Automatic Upgrades for Unity Catalog managed tables, a capability that helps tables adopt recommended, generally available features automatically after Databricks verifies workload compatibility.
Whatโs new
- Best-practice features applied automatically: Automatic Upgrades brings recommended table features to Unity Catalog managed tables without requiring manual ALTER TABLE work across large numbers of tables.
- Compatibility is checked before anything changes: Databricks uses an observation window to monitor access patterns and only turns on features after verifying that all accessing workloads are compatible.
- Focused on managed tables: The feature applies to Unity Catalog managed tables, which Databricks already recommends for lower cost, better performance, secure external access, and automatic platform improvements.
- Teams stay in control: If a feature is turned on automatically, admins can still turn it off on an individual table, and Automatic Upgrades will not turn that feature back on for that table later.
- Designed to improve performance, reliability, and interoperability over time: Supported features include items such as row tracking, Checkpoint V2, automatic liquid clustering for new tables, catalog commits, and Parquet v2, depending on eligibility and rollout stage.
Databricks also positions this as a simpler alternative to manual feature rollout. Instead of auditing thousands of tables by hand, teams can let Databricks observe usage, verify compatibility, and apply upgrades in the background using serverless compute at no charge for the upgrade process itself.
๐ Read the full post here
