Description
Databricks Lakebase lets you branch your database like code. A new branch of your serverless Postgres database inherits both the schema and the data of its parent in seconds, so you can test changes on a production-like database without touching production.
⏰ Chapters
00:00 - Database Branching in Lakebase Serverless Postgres
00:42 - How Branching Works: Copy-on-Write, Compute and No Merge
01:45 - Create a Database Branch with the Databricks CLI
02:56 - Test a Schema Migration on a Branch and Review Schema Diff
03:59 - Apply the Tested Migration to Production and Reset
04:48 - Point-in-Time Restore After an Accidental DELETE
05:38 - Recover Production with pg_dump and Clean Up Branches
🌿 How Lakebase branching works
A branch shares every storage page with its parent through copy-on-write instead of copying the data, so creating one takes seconds no matter how big the database is. It stores only the pages you change and runs on its own compute that scales to zero when idle, which lets developers, CI pipelines, and AI coding agents each work on their own branch.
🛠️ What you'll see in the demo
Using a coffee shop database with 2,500 customers and 120,000 orders:
• Creating a branch with the Databricks CLI in about 4 seconds, with a TTL that deletes it after 1 day
• Connecting with psql and testing a schema migration on the branch while production stays untouched
• Reviewing the change with schema diff, which compares the schema of the branch and its parent, not the rows
• Applying the same tested migration file to production, because Lakebase does not merge data between branches the way Git merges code
• Resetting the branch from its parent to start the next change with fresh production data
• Recovering from an accidental DELETE with point-in-time restore, which creates a new branch from the 7-day history window instead of overwriting production
• Copying the orders back with pg_dump and letting idle branches scale to zero
Whether you're a developer testing a risky migration, a platform engineer rolling out database branching to your team, or a team lead giving AI coding agents a safe copy of production, Lakebase branching lets you change a real copy of production, apply it with the same migration file, and get your data back in minutes if something goes wrong.
🔗 Resources
📘 Lakebase branches documentation: https://docs.databricks.com/aws/en/oltp/projects/branches
🚀 Databricks Lakebase: https://www.databricks.com/product/lakebase
🤝 Stay connected
Website: https://databricks.com
Twitter: https://x.com/databricks/
LinkedIn: https://www.linkedin.com/company/databricks/
Instagram: https://www.instagram.com/databricksinc/
Facebook: https://www.facebook.com/databricksinc/