Skip to main content

Lakebase Postgres

Lakebase Postgres is a fully managed, cloud-native PostgreSQL database that brings online transaction processing (OLTP) capabilities to the Lakehouse.

Choose your version

Two versions of Lakebase Postgres are available, both in Public Preview.

Version

Description

Lakebase (Provisioned Preview)

Provisioned compute version. Fully managed PostgreSQL, supported on AWS and Azure. Organizes resources as database instances. You manually scale compute capacity based on your workload needs. Supports high availability (HA) and Databricks integrations not yet available in the Lakebase Autoscaling Preview. See What is Lakebase (Provisioned)? for details.

Lakebase (Autoscaling Preview)

Modern database capabilities with autoscaling. The new version of Lakebase. Fully managed PostgreSQL, supported on AWS. Organizes resources as projects that contain branches, computes, and databases. Features autoscaling compute, branching, scale-to-zero, and instant restore. See What is Lakebase (Autoscaling Preview)? for details.

Lakebase Autoscaling Preview access

If you don't see Lakebase Postgres (Autoscaling Preview) in your workspace, workspace admins can enable it through Settings > Workspace settings > Preview features. For information about enabling Preview features, see Manage previews.

What's new in Lakebase Autoscaling Preview

Lakebase Autoscaling Preview introduces modern database capabilities designed for flexible development and cost-efficient operations:

  • New Lakebase App: Redesigned interface with an intuitive project-based structure for managing database resources
  • Autoscaling: Compute resources automatically adjust based on workload demand. Learn more: Autoscaling
  • Scale-to-zero: Computes automatically suspend after inactivity, waking up in seconds when needed. Learn more: Scale to zero
  • Projects and branches: Create instant, isolated copies of your database using copy-on-write technology. Learn more: Branches
  • Instant restore: Restore your database to any point in time within your configured restore window (2-35 days). Learn more: Point-in-time restore
  • Read replicas: Create read-only compute endpoints that share the same storage layer, enabling instant creation without data duplication. Learn more: Read replicas
  • Data API: PostgREST-compatible RESTful interface for direct HTTP access to your database. Learn more: Data API

Feature comparison

The following table compares capabilities in the Lakebase Provisioned Preview and Autoscaling Preview.

note

The Autoscaling Preview is the new version of Lakebase and is progressing toward feature parity with the Provisioned Preview.

Feature

Provisioned

Autoscaling Preview

Core capabilities

Provisioned (fixed-size) compute

Autoscaling

Scale to zero

Branching

Instant restore

Point-in-time restore

Read replicas

High availability (readable secondaries)

Data integrations

Unity Catalog registration

Reverse ETL

Application integrations

Databricks Apps, Feature Store, Notebooks, Databricks SQL Editor

Access control

Postgres roles and permissions

Workspace ACLs

Developer tools

Infrastructure as code (Asset Bundles, Terraform)

Programmatic access (REST API, CLI, SDKs)

PostgREST API support

✓ (Private Preview)

✓ (Public Preview, PostgREST-compatible Data API)

Cost management

Billing tags and budget policies

Choosing the right version

Both versions of Lakebase are suitable for production workloads. Choose based on your feature requirements:

Choose Lakebase Provisioned Preview if you want:

  • Provisioned compute that you manually scale
  • Azure support (Autoscaling Preview is AWS only)
  • Currently available on additional AWS regions
  • Databricks integrations not yet supported on the Autoscaling Preview

Choose Lakebase Autoscaling Preview if you want:

  • Automatic scaling based on workload demand
  • Scale-to-zero for cost optimization
  • Branch-based development workflows
  • Instant point-in-time restore

For detailed feature differences including platform integrations, see the Feature comparison table above and Lakebase Autoscaling Preview limitations.