Databricks says Lakebase Postgres restores 100 TB databases in seconds via branch-based restore

Databricks says its Lakebase Postgres service restores a 100 TB database in seconds using branch-based restores.

According to the company, a restore is a metadata operation that creates a branch at a chosen point in time by referencing immutable history already held in decoupled storage, without copying data or replaying WAL. Databricks says recovery time stays near-instant regardless of database size. The claim comes from a Databricks post on X; no independent measurements are reported.