GROUPING SETS / ROLLUP / CUBE

Database	Status	Syntax Family	Min Version
PostgreSQL	full	sql-standard	9.5
MySQL	partial	rollup-only	-
MariaDB	partial	rollup-only	-
Oracle	full	sql-standard	-
SQL Server	full	sql-standard	2008
SQLite	none	-	-
Redshift	full	sql-standard	-
DB2	full	sql-standard	-
Snowflake	full	sql-standard	-
BigQuery	full	sql-standard	-

Compare Databases

Database Details

PostgreSQL full

SELECT
  region,
  product,
  SUM(sales) AS total_sales
FROM sales_data
GROUP BY GROUPING SETS (
  (region, product),
  (region),
  (product),
  ()
)
ORDER BY region NULLS LAST, product NULLS LAST;

ROLLUP, CUBE, and GROUPING SETS were all added in 9.5.
Multiple grouping specifications can be combined in one GROUP BY clause.

Docs: GROUPING SETS, CUBE, and ROLLUP

MySQL partial

SELECT region, product, SUM(sales) AS total_sales
FROM sales_data
GROUP BY region, product WITH ROLLUP;

ROLLUP only; no CUBE or GROUPING SETS.
GROUPING() was added in 8.0.

Docs: GROUP BY Modifiers

MariaDB partial

SELECT region, product, SUM(sales) AS total_sales
FROM sales_data
GROUP BY region, product WITH ROLLUP;

ROLLUP only, with MySQL's WITH ROLLUP trailing-modifier spelling. No CUBE or GROUPING SETS.

Docs: GROUP BY WITH ROLLUP (MariaDB)

Oracle full

SELECT
  region,
  product,
  SUM(sales) AS total_sales
FROM sales_data
GROUP BY GROUPING SETS (
  (region, product),
  (region),
  (product),
  ()
)
ORDER BY region NULLS LAST, product NULLS LAST;

Oracle shipped ROLLUP and CUBE first among major vendors (8i).
Supports composite columns and concatenated grouping sets.

Docs: ROLLUP, CUBE, GROUPING SETS (Oracle)

SQL Server full

SELECT
  region,
  product,
  SUM(sales) AS total_sales
FROM sales_data
GROUP BY GROUPING SETS (
  (region, product),
  (region),
  (product),
  ()
)
ORDER BY region, product;

All three added in 2008.
GROUPING_ID() returns a bitmap of which columns a given row aggregates over — useful when multiple groupings are combined.

Docs: GROUP BY (Transact-SQL)

SQLite none

Not supported. Simulate by UNION ALL of separate GROUP BY queries.

Docs: SQLite SELECT

Redshift full

SELECT
  region,
  product,
  SUM(sales) AS total_sales
FROM sales_data
GROUP BY GROUPING SETS (
  (region, product),
  (region),
  (product),
  ()
)
ORDER BY region NULLS LAST, product NULLS LAST;

Docs: GROUP BY (Redshift)

DB2 full

SELECT
  region,
  product,
  SUM(sales) AS total_sales
FROM sales_data
GROUP BY GROUPING SETS (
  (region, product),
  (region),
  (product),
  ()
)
ORDER BY region, product;

Docs: Grouping Sets, Rollup, and Cube (IBM Db2)

Snowflake full

SELECT
  region,
  product,
  SUM(sales) AS total_sales
FROM sales_data
GROUP BY GROUPING SETS (
  (region, product),
  (region),
  (product),
  ()
)
ORDER BY region NULLS LAST, product NULLS LAST;

Docs: ROLLUP, CUBE, and GROUPING SETS (Snowflake)

BigQuery full

SELECT
  region,
  product,
  SUM(sales) AS total_sales
FROM sales_data
GROUP BY GROUPING SETS (
  (region, product),
  (region),
  (product),
  ()
)
ORDER BY region NULLS LAST, product NULLS LAST;

ROLLUP was there from the start of GoogleSQL; CUBE and GROUPING SETS arrived in 2022.

Docs: GROUP BY Clause (BigQuery)