Does Databricks have an operational database?

Summary

• Yes. Databricks' operational database is Lakebase — a fully managed, serverless, PostgreSQL-compatible OLTP database built into the Databricks Data Intelligence Platform.
• Separation of storage and compute. Lakebase separates stateless Postgres compute from durable storage, which enables scale-to-zero (idle compute suspends automatically) and independent scaling.
• Built for transactions. Applications and AI agents connect to Lakebase for low-latency transactional reads and writes with transactional consistency.
• Instant branching and recovery. Create isolated, zero-copy branches of your database in seconds for safe testing, with git-like branching, time travel, and point-in-time recovery.
• Governed and lakehouse-connected. Lakebase integrates natively with Unity Catalog, and synced tables serve governed lakehouse data into Postgres for low-latency reads.

Does Databricks have an operational database?

Yes. Databricks' operational database is Lakebase, a fully managed, serverless, PostgreSQL-compatible OLTP (online transactional processing) database that is built into the Databricks Data Intelligence Platform. It is designed for low-latency operational and transactional workloads that run directly on the same data powering your analytical and AI use cases, and it is generally available. See the announcement in Databricks Lakebase is generally available.

Why Databricks Lakebase

• Fully managed, serverless Postgres. Lakebase is PostgreSQL-compatible, so you can use the Postgres tools and drivers you already know while Databricks manages the database for you. See Lakebase projects.
• Separation of storage and compute. Lakebase separates stateless Postgres compute from a durable storage layer built on safekeepers, pageservers, and cloud object storage. This architecture is what enables scale-to-zero and elastic scaling. See Lakebase core concepts.
• Scale-to-zero. Compute automatically adjusts to match traffic and suspends completely when idle, so you minimize cost during quiet periods.
• Low-latency transactional reads and writes. Lakebase is built for fast lookup-style queries and transactional consistency, so applications and AI agents can read and write operational data with low latency.
• Instant branching and recovery. Lakebase supports instant, zero-copy database branching — create isolated clones of production data in seconds for risk-free testing and development — along with git-like branching, time travel, and point-in-time recovery.
• Native Unity Catalog governance. Lakebase integrates natively with Unity Catalog, so the same governance model that applies to Unity Catalog managed tables applies to Lakebase data, and Lakebase data can be registered in Unity Catalog for federated queries and cross-source analysis.
• Synced tables to serve lakehouse data. Synced tables bring Unity Catalog data into Lakebase Postgres for low-latency transactional reads. Continuous mode keeps data within seconds of the source, while Triggered mode uses scheduled incremental updates to balance freshness and cost. See Serve lakehouse data with synced tables.

Getting started

• Read Lakebase projects to understand the managed Postgres experience.
• Review Lakebase core concepts for the storage-and-compute architecture, branching, and scale-to-zero.
• Set up synced tables to serve governed lakehouse data into Postgres for your applications.

FAQs

What is Lakebase?

Lakebase is Databricks' fully managed, serverless, PostgreSQL-compatible operational (OLTP) database, built into the Databricks Data Intelligence Platform for low-latency transactional workloads.

Is Lakebase compatible with PostgreSQL?

Yes. Lakebase is PostgreSQL-compatible, so you can use standard Postgres tools, drivers, and SQL.

How does Lakebase connect to the lakehouse?

Lakebase integrates natively with Unity Catalog. Synced tables bring governed lakehouse data into Postgres for low-latency reads, and Lakebase data can be registered in Unity Catalog for federated queries.

What does scale-to-zero mean for Lakebase?

Because Lakebase separates compute from storage, compute suspends automatically when the database is idle and scales back up with traffic, so you pay for compute only when it is in use.