12.20.2 GROUP BY Modifiers

The GROUP BY clause permits a WITH ROLLUP modifier that causes summary output to include extra rows that represent higher-level (that is, super-aggregate) summary operations. ROLLUP thus enables you to answer questions at multiple levels of analysis with a single query. For example, ROLLUP can be used to provide support for OLAP (Online Analytical Processing) operations.

Suppose that a sales table has year, country, product, and profit columns for recording sales profitability:

To summarize table contents per year, use a simple GROUP BY like this:

The output shows the total (aggregate) profit for each year. To also determine the total profit summed over all years, you must add up the individual values yourself or run an additional query. Or you can use ROLLUP, which provides both levels of analysis with a single query. Adding a WITH ROLLUP modifier to the GROUP BY clause causes the query to produce another (super-aggregate) row that shows the grand total over all year values:

The NULL value in the year column identifies the grand total super-aggregate line.

ROLLUP has a more complex effect when there are multiple GROUP BY columns. In this case, each time there is a change in value in any but the last grouping column, the query produces an extra super-aggregate summary row.

For example, without ROLLUP, a summary of the sales table based on year, country, and product might look like this, where the output indicates summary values only at the year/country/product level of analysis:

With ROLLUP added, the query produces several extra rows:

Now the output includes summary information at four levels of analysis, not just one:

The NULL indicators in each super-aggregate row are produced when the row is sent to the client. The server looks at the columns named in the GROUP BY clause following the leftmost one that has changed value. For any column in the result set with a name that matches any of those names, its value is set to NULL. (If you specify grouping columns by column position, the server identifies which columns to set to NULL by position.)

Because the NULL values in the super-aggregate rows are placed into the result set at such a late stage in query processing, you can test them as NULL values only in the select list or HAVING clause. You cannot test them as NULL values in join conditions or the WHERE clause to determine which rows to select. For example, you cannot add WHERE product IS NULL to the query to eliminate from the output all but the super-aggregate rows.

The NULL values do appear as NULL on the client side and can be tested as such using any MySQL client programming interface. However, at this point, you cannot distinguish whether a NULL represents a regular grouped value or a super-aggregate value. To test the distinction, use the GROUPING() function, described later.

Previously, MySQL did not allow the use of DISTINCT or ORDER BY in a query having a WITH ROLLUP option. This restriction is lifted in MySQL 8.0.12 and later. (Bug #87450, Bug #86311, Bug #26640100, Bug #26073513)

For GROUP BY ... WITH ROLLUP queries, to test whether NULL values in the result represent super-aggregate values, the GROUPING() function is available for use in the select list, HAVING clause, and (as of MySQL 8.0.12) ORDER BY clause. For example, GROUPING(year) returns 1 when NULL in the year column occurs in a super-aggregate row, and 0 otherwise. Similarly, GROUPING(country) and GROUPING(product) return 1 for super-aggregate NULL values in the country and product columns, respectively:

Instead of displaying the GROUPING() results directly, you can use GROUPING() to substitute labels for super-aggregate NULL values:

With multiple expression arguments, GROUPING() returns a result representing a bitmask the combines the results for each expression, with the lowest-order bit corresponding to the result for the rightmost expression. For example, GROUPING(year, country, product) is evaluated like this:

  result for GROUPING(product)
+ result for GROUPING(country) << 1
+ result for GROUPING(year) << 2

The result of such a GROUPING() is nonzero if any of the expressions represents a super-aggregate NULL, so you can return only the super-aggregate rows and filter out the regular grouped rows like this:

The sales table contains no NULL values, so all NULL values in a ROLLUP result represent super-aggregate values. When the data set contains NULL values, ROLLUP summaries may contain NULL values not only in super-aggregate rows, but also in regular grouped rows. GROUPING() enables these to be distinguished. Suppose that table t1 contains a simple data set with two grouping factors for a set of quantity values, where NULL indicates something like “other” or “unknown”:

A simple ROLLUP operation produces these results, in which it is not so easy to distinguish NULL values in super-aggregate rows from NULL values in regular grouped rows:

Using GROUPING() to substitute labels for the super-aggregate NULL values makes the result easier to interpret: