Concept card

Databricks Lakebase — serverless Postgres on lakehouse storage, and what it means in SAP Databricks

As of 2026-11-04

Edited by Cédric Mary, publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

What is Databricks Lakebase?

A Lakebase, in Databricks' own definition, is a transactional database whose compute is separate from storage, with the data kept in open formats in cloud object storage and Postgres running as serverless compute on top. Databricks claims scale to zero, instant branching and shared OLTP and OLAP storage. SAP announced Lakebase for SAP Databricks in SAP Business Data Cloud on 27 and 29 Oct 2026; the SAP sources give no availability date.

What a Lakebase is

A Lakebase is the term Databricks uses for a transactional database rebuilt on a data-lake design. The Databricks blog "A New Era of Databases: Lakebase" (published 12 Jun 2025, updated 24 Feb 2026) defines it as an open architecture that separates compute from storage and places the database's data directly in low-cost cloud object storage, in open formats, while the transactional compute layer runs independently on top. The engine named in the blog is Postgres, delivered as a fully managed, serverless service. Everything below is Databricks' own description of its product category; it is a vendor view, not an independent benchmark.

The problems Databricks says it solves

The blog lists three limits of traditional databases. First, fragile and costly operations: compute and storage are bundled, so teams provision for peak load, pay for idle capacity, and routine tasks such as snapshots or a GDPR clean-up query can strain the whole system. Second, a clunky developer experience: a git branch of code takes under a second, while an isolated copy of a production database takes minutes to hours, which matters more now that AI agents want temporary environments. Third, vendor lock-in: data can only be reached through the database engine itself, so migrations are feared projects.

Why it matters

Agents need somewhere to keep state, queues and memory with transactional guarantees; a Lakebase is Databricks' answer, and SAP has announced it for SAP Databricks.
The three-generation story is a vendor argument; knowing whose framing you are hearing keeps a consultant's advice neutral.
Nothing in the SAP sources gives a date, price or limit for Lakebase in SAP Databricks, so a client plan must not depend on it yet.

Key points

Databricks defines a Lakebase as an open architecture: compute separated from storage, data in open formats in cloud object storage (Databricks blog, 12 Jun 2025, updated 24 Feb 2026).
The engine named is Postgres, delivered as managed, serverless compute that scales with demand and down when idle.
Databricks claims scale to zero in under a second and copies of even petabyte-scale databases in seconds: vendor claims, not benchmarks.
Transactional and analytical workloads share the storage layer, so analytics and AI run on transactional data without moving it (Databricks wording).
The three-generation framing (monolith, proprietary loose coupling, Lakebase) is Databricks' own argument.
SAP announced Lakebase for SAP Databricks in SAP BDC (SAP News, 27 and 29 Oct 2026).
The SAP TechEd 2026 Innovation News Guide lists the SAP Databricks expansion as announced, with no date (SAP, 27 Oct 2026).
Using Lakebase for agent state next to SAP data products is our inference, not an SAP statement.

Terms used on this page

Lakebase: Databricks' term for a transactional database with compute separate from storage and data in open formats in cloud object storage.
OLTP: Online transaction processing: many small reads and writes with transactional guarantees.
OLAP: Online analytical processing: large scans and aggregations over history.
Serverless Postgres: Postgres compute that a provider manages and scales with demand, down to zero when idle.
Scale to zero: Releasing compute when there is no load so that no idle machine is paid for.
Branching: Creating an isolated, writable copy of a database for development, like a git branch.
Open format: A documented storage format that several engines can read, so the data is not tied to one database engine.
SAP Databricks: Databricks capabilities delivered inside SAP Business Data Cloud, where SAP announced Lakebase on 27 Oct 2026.

Sources

Full card with the Consultant plan — €19.90/month. What the full card adds: the full decision framework · the SAP vs Snowflake / Databricks / Fabric comparison · the common pitfalls and their fix · the cheat sheet · the architecture schemas · the code blocks · the facts worth quoting.

Open in the app →

Open missions

SAP analytics jobs and contracts · freelance SAP contracts.