# Databricks SQL

> Managed SQL warehouse over open lakehouse tables

Databricks SQL is the warehousing surface of the Databricks platform, running BI and analytical SQL on the vectorized Photon engine against Delta Lake and Iceberg tables governed by Unity Catalog.

Website: https://www.databricks.com/product/databricks-sql

## Use it when

- Spark, ML, and BI already share the platform and the same tables should serve all three without a second copy.
- Open storage is a requirement: Delta Lake and Iceberg files stay in your object storage, readable by any engine.
- Unity Catalog governance (row filters, column masks, lineage, audit) should span the whole lakehouse, not just the warehouse.
- Serverless warehouses that start in seconds and stop when idle fit your usage pattern.

## Think twice when

- You only need a warehouse; against Snowflake or BigQuery you are buying a platform whose surface area extends well past SQL.
- Cost needs modeling: DBU rates differ across Classic, Pro, and Serverless, and the first two add the cloud VM bill.
- The team has no Spark gravity; the openness trade pays off most when other engines and workloads share the tables.

## How it runs

Managed on AWS, Azure, and Google Cloud. SQL warehouses run the vectorized Photon engine against Delta and Iceberg tables in your object storage, governed by Unity Catalog. Classic and Pro compute lives in your cloud account; Serverless lives in Databricks' and bundles infrastructure into the DBU rate. Streaming tables ingest from Kafka, Kinesis, and Pub/Sub.

## Details

### General

| Attribute | Value |
| --- | --- |
| License | Commercial |
| Deployment | Managed, Serverless |
| Operational complexity | Medium |
| Pricing | Free, Subscription |
| Documentation | https://docs.databricks.com/aws/en/sql/ |

### Data Warehouses

| Attribute | Value |
| --- | --- |
| Where it runs | AWS, Azure and Google Cloud |
| Compute and storage | MPP over object storage: classic and pro compute in your cloud account, serverless in Databricks' |
| Storage format | Open: Delta Lake and managed or foreign Iceberg in your object storage |
| SQL dialect | ANSI SQL (Spark SQL) |
| Streaming ingest | Native: streaming tables from Kafka, Kinesis, Pub/Sub and cloud storage |
| Concurrent users | One cluster per 10 concurrent queries, autoscaled; serverless adds intelligent workload management |
| Acceleration | Photon vectorized engine, predictive I/O, disk cache, materialized views, predictive optimization |
| Billing unit | DBU-second per warehouse; classic and pro add the cloud VM bill, serverless bundles it |
| Time travel | Delta time travel: 7 days of data files by default, 30 days of log, configurable |
| Access control | Unity Catalog: GRANT on a three-level namespace, row filters and column masks, tag-based ABAC, lineage and audit |

## Capabilities

- [Data Warehouses](https://matca.io/capabilities/data-warehouse)

---

Source: https://matca.io/databricks-sql
Last updated: 2026-08-31T11:40:49.596Z