/
New feature GA

Serverless SQL warehouses

Serverless SQL warehouses is a Databricks compute / bi capability, introduced August 2021.

The SQL warehouse type that runs in the Databricks-managed serverless compute plane rather than your cloud account, supporting Photon, Predictive IO and Intelligent Workload Management, and the default type in the UI where serverless is available.

  • Its launch note handed everything that came before it a retronym - the SQL endpoints living in your own AWS account were, from that day, now called Classic SQL endpoints.
  • On the day it went GA, Databricks stopped creating a serverless starter SQL warehouse for you.
  • You do not get to pick the chip - Databricks might change the CPU architecture behind a serverless SQL warehouse between restarts, without notice.

Limitations: Cluster policies (including spot instance policies) are not supported, customer-managed VPCs do not apply, customer-managed keys are not used for EBS storage encryption, and a workspace with an external Hive legacy metastore or workspace-level Hive metastore credentials cannot deploy them.

Open in REbricked →
Category
Compute / BI
Introduced
August 2021
Also known as
Serverless SQL warehouse, Serverless SQL endpoints, Serverless SQL, Databricks SQL Serverless, DBSQL Serverless
Verified
2026-09-27

Sources

Related in Compute / BI