Skip to main content

Druid SQL overview


Apache Druid supports two query languages: Druid SQL and native queries. This document describes the SQL language.

You can query data in Druid datasources using Druid SQL. Druid translates SQL queries into its native query language. To learn about translation and how to get the best performance from Druid SQL, see SQL query translation.

Druid SQL planning occurs on the Broker. Set Broker runtime properties to configure the query plan and JDBC querying.

For information on permissions needed to make SQL queries, see Defining SQL permissions.

This topic introduces Druid SQL syntax. For more information and SQL querying options see:

  • Data types for a list of supported data types for Druid columns.
  • Aggregation functions for a list of aggregation functions available for Druid SQL SELECT statements.
  • Scalar functions for Druid SQL scalar functions including numeric and string functions, IP address functions, Sketch functions, and more.
  • SQL multi-value string functions for operations you can perform on string dimensions containing multiple values.
  • Query translation for information about how Druid translates SQL queries to native queries before running them.

For information about APIs, see:


Druid SQL supports SELECT queries with the following structure:

[ WITH tableName [ ( column1, column2, ... ) ] AS ( query ) ]
SELECT [ ALL | DISTINCT ] { * | exprs }
FROM { <table> | (<subquery>) | <o1> [ INNER | LEFT ] JOIN <o2> ON condition }
[PIVOT (aggregation_function(column_to_aggregate) FOR column_with_values_to_pivot IN (pivoted_column1 [, pivoted_column2 ...]))]
[UNPIVOT (values_column FOR names_column IN (unpivoted_column1 [, unpivoted_column2 ... ]))]
[ CROSS JOIN UNNEST(source_expression) as table_alias_name(column_alias_name) ]
[ WHERE expr ]
[ GROUP BY [ exprs | GROUPING SETS ( (exprs), ... ) | ROLLUP (exprs) | CUBE (exprs) ] ]
[ HAVING expr ]
[ ORDER BY expr [ ASC | DESC ], expr [ ASC | DESC ], ... ]
[ LIMIT limit ]
[ OFFSET offset ]
[ UNION ALL <another query> ]


The FROM clause can refer to any of the following:

  • Table datasources from the druid schema. This is the default schema, so Druid table datasources can be referenced as either druid.dataSourceName or simply dataSourceName.
  • Lookups from the lookup schema, for example lookup.countries. Note that lookups can also be queried using the LOOKUP function.
  • Subqueries.
  • Joins between anything in this list, except between native datasources (table, lookup, query) and system tables. The join condition must be an equality between expressions from the left- and right-hand side of the join.
  • Metadata tables from the INFORMATION_SCHEMA or sys schemas. Unlike the other options for the FROM clause, metadata tables are not considered datasources. They exist only in the SQL layer.

For more information about table, lookup, query, and join datasources, refer to the Datasources documentation.



The PIVOT operator is an experimental feature.

The PIVOT operator carries out an aggregation and transforms rows into columns in the output.

The following is the general syntax for the PIVOT operator. Note that the PIVOT operator is enclosed in parentheses and forms part of the FROM clause of the query.

PIVOT (aggregation_function(column_to_aggregate)
FOR column_with_values_to_pivot
IN (pivoted_column1 [, pivoted_column2 ...])

PIVOT syntax parameters:

  • aggregation_function: An aggregation function, such as SUM, COUNT, MIN, MAX, or AVG.
  • column_to_aggregate: The source column to be aggregated.
  • column_with_values_to_pivot: The column that contains values for the pivoted column names.
  • pivoted_columnN: The list of values to pivot into headers in the output.

The following example demonstrates how to transform cityName values into column headers ba_sum_deleted and ny_sum_deleted:

SELECT user, channel, ba_sum_deleted, ny_sum_deleted
FROM "wikipedia"
PIVOT (SUM(deleted) AS "sum_deleted" FOR "cityName" IN ( 'Buenos Aires' AS ba, 'New York' AS ny))
WHERE ba_sum_deleted IS NOT NULL OR ny_sum_deleted IS NOT NULL
View results



The UNPIVOT operator is an experimental feature.

The UNPIVOT operator transforms existing column values into rows. Note that UNPIVOT isn't the exact reverse operation of PIVOT. The PIVOT operator carries out an aggregation and merges rows as needed. UNPIVOT doesn't reproduce the original rows that have been merged.

The following is the general syntax for the UNPIVOT operator. Note that the UNPIVOT operator is enclosed in parentheses and forms part of the FROM clause of the query.

UNPIVOT (values_column 
FOR names_column
IN (unpivoted_column1 [, unpivoted_column2 ... ])

UNPIVOT syntax parameters:

  • values_column: The column that contains the values of the unpivoted columns.
  • names_column: The column that contains the names of the unpivoted columns.
  • unpivoted_columnN: The list of columns to transform into rows in the output.

The following example demonstrates how to transform the columns added and deleted into row values that correspond to a particular channel:

SELECT channel, user, action, SUM(changes) AS total_changes
FROM "wikipedia"
UNPIVOT ( changes FOR action IN ("added", "deleted") )
WHERE channel LIKE '#ar%'
GROUP BY channel, user, action
View results


The UNNEST clause unnests ARRAY typed values. The source for UNNEST can be an array type column, or an input that's been transformed into an array, such as with helper functions like MV_TO_ARRAY or ARRAY.

The following is the general syntax for UNNEST, specifically a query that returns the column that gets unnested:

SELECT column_alias_name FROM datasource CROSS JOIN UNNEST(source_expression1) AS table_alias_name1(column_alias_name1) CROSS JOIN UNNEST(source_expression2) AS table_alias_name2(column_alias_name2) ...
  • The datasource for UNNEST can be any Druid datasource, such as the following:
    • A table, such as FROM a_table.
    • A subset of a table based on a query, a filter, or a JOIN. For example, FROM (SELECT columnA,columnB,columnC from a_table).
  • The source_expression for the UNNEST function must be an array and can come from any expression. UNNEST works directly on Druid ARRAY typed columns. If the column you are unnesting is a multi-value VARCHAR, you must specify MV_TO_ARRAY(dimension) to convert it to an ARRAY type. You can also specify any expression that has an SQL array datatype. For example, you can call UNNEST on the following:
    • ARRAY[dim1,dim2] if you want to make an array out of two dimensions.
    • ARRAY_CONCAT(dim1,dim2) if you want to concatenate two multi-value dimensions.
  • The AS table_alias_name(column_alias_name) clause is not required but is highly recommended. Use it to specify the output, which can be an existing column or a new one. Replace table_alias_name and column_alias_name with a table and column name you want to alias the unnested results to. If you don't provide this, Druid uses a nondescriptive name, such as EXPR$0.

Keep the following things in mind when writing your query:

  • You can unnest multiple source expressions in a single query.
  • Notice the CROSS JOIN between the datasource and the UNNEST function. This is needed in most cases of the UNNEST function. Specifically, it is not needed when you're unnesting an inline array since the array itself is the datasource.
  • If you view the native explanation of a SQL UNNEST, you'll notice that Druid uses j0.unnest as a virtual column to perform the unnest. An underscore is added for each unnest, so you may notice virtual columns named _j0.unnest or __j0.unnest.
  • UNNEST preserves the ordering of the source array that is being unnested.

For examples, see the Unnest arrays tutorial.

The UNNEST function has the following limitations:

  • The function does not remove any duplicates or nulls in an array. Nulls will be treated as any other value in an array. If there are multiple nulls within the array, a record corresponding to each of the nulls gets created.
  • Arrays of complex objects inside complex JSON types are not supported.

UNNEST is the SQL equivalent of the unnest datasource.


The WHERE clause refers to columns in the FROM table, and will be translated to native filters. The WHERE clause can also reference a subquery, like WHERE col1 IN (SELECT foo FROM ...). Queries like this are executed as a join on the subquery, described in the Query translation section.

Strings and numbers can be compared in the WHERE clause of a SQL query through implicit type conversion. For example, you can evaluate WHERE stringDim = 1 for a string-typed dimension named stringDim. However, for optimal performance, you should explicitly cast the reference number as a string when comparing against a string dimension:

WHERE stringDim = '1'

Similarly, if you compare a string-typed dimension with reference to an array of numbers, cast the numbers to strings:

WHERE stringDim IN ('1', '2', '3')

Note that explicit type casting does not lead to significant performance improvement when comparing strings and numbers involving numeric dimensions since numeric dimensions are not indexed.


The GROUP BY clause refers to columns in the FROM table. Using GROUP BY, DISTINCT, or any aggregation functions will trigger an aggregation query using one of Druid's three native aggregation query types. GROUP BY can refer to an expression or a select clause ordinal position (like GROUP BY 2 to group by the second selected column).

The GROUP BY clause can also refer to multiple grouping sets in three ways. The most flexible is GROUP BY GROUPING SETS, for example GROUP BY GROUPING SETS ( (country, city), () ). This example is equivalent to a GROUP BY country, city followed by GROUP BY () (a grand total). With GROUPING SETS, the underlying data is only scanned one time, leading to better efficiency. Second, GROUP BY ROLLUP computes a grouping set for each level of the grouping expressions. For example GROUP BY ROLLUP (country, city) is equivalent to GROUP BY GROUPING SETS ( (country, city), (country), () ) and will produce grouped rows for each country / city pair, along with subtotals for each country, along with a grand total. Finally, GROUP BY CUBE computes a grouping set for each combination of grouping expressions. For example, GROUP BY CUBE (country, city) is equivalent to GROUP BY GROUPING SETS ( (country, city), (country), (city), () ).

Grouping columns that do not apply to a particular row will contain NULL. For example, when computing GROUP BY GROUPING SETS ( (country, city), () ), the grand total row corresponding to () will have NULL for the "country" and "city" columns. Column may also be NULL if it was NULL in the data itself. To differentiate such rows, you can use GROUPING aggregation.

When using GROUP BY GROUPING SETS, GROUP BY ROLLUP, or GROUP BY CUBE, be aware that results may not be generated in the order that you specify your grouping sets in the query. If you need results to be generated in a particular order, use the ORDER BY clause.


The HAVING clause refers to columns that are present after execution of GROUP BY. It can be used to filter on either grouping expressions or aggregated values. It can only be used together with GROUP BY.


The ORDER BY clause refers to columns that are present after execution of GROUP BY. It can be used to order the results based on either grouping expressions or aggregated values. ORDER BY can refer to an expression or a select clause ordinal position (like ORDER BY 2 to order by the second selected column). For non-aggregation queries, ORDER BY can only order by the __time column. For aggregation queries, ORDER BY can order by any column.


The LIMIT clause limits the number of rows returned. In some situations Druid will push down this limit to data servers, which boosts performance. Limits are always pushed down for queries that run with the native Scan or TopN query types. With the native GroupBy query type, it is pushed down when ordering on a column that you are grouping by. If you notice that adding a limit doesn't change performance very much, then it's possible that Druid wasn't able to push down the limit for your query.


The OFFSET clause skips a certain number of rows when returning results.

If both LIMIT and OFFSET are provided, then OFFSET will be applied first, followed by LIMIT. For example, using LIMIT 100 OFFSET 10 will return 100 rows, starting from row number 10.

Together, LIMIT and OFFSET can be used to implement pagination. However, note that if the underlying datasource is modified between page fetches, then the different pages will not necessarily align with each other.

There are two important factors that can affect the performance of queries that use OFFSET:

  • Skipped rows still need to be generated internally and then discarded, meaning that raising offsets to high values can cause queries to use additional resources.
  • OFFSET is only supported by the Scan and GroupBy native query types. Therefore, a query with OFFSET will use one of those two types, even if it might otherwise have run as a Timeseries or TopN. Switching query engines in this way can affect performance.


The UNION ALL operator fuses multiple queries together. Druid SQL supports the UNION ALL operator in two situations: top-level and table-level, as described below. Queries that use UNION ALL in any other way will fail.


In top-level queries, you can use UNION ALL at the very top outer layer of the query - not in a subquery, and not in the FROM clause. The underlying queries run sequentially. Druid concatenates their results so that they appear one after the other.

For example:

SELECT COUNT(*) FROM tbl WHERE my_column = 'value1'
SELECT COUNT(*) FROM tbl WHERE my_column = 'value2'

Certain limitations apply when you use a top-level UNION ALL. For all top-level UNION ALL queries, you can't apply a GROUP BY, ORDER BY, or any other operator to the results of the query. For any top-level UNION ALL that uses the MSQ task engine, the SQL planner attempts to plan the top-level UNION ALL as a table-level UNION ALL. Because of this, UNION ALL queries that use the MSQ task engine always behave the same as table-level UNION ALL queries. They have the same characteristics and limitations. If the planner can't plan the query as a table-level UNION ALL, the query fails.


In table-level queries, you must use UNION ALL in a subquery in the FROM clause, and create the lower-level subqueries that are inputs to the UNION ALL operator as simple table SELECTs. You can't use features like expressions, column aliasing, JOIN, GROUP BY, or ORDER BY in table-level queries.

The query runs natively using a union datasource.

At table-level queries, you must select the same columns from each table in the same order, and those columns must either have the same types, or types that can be implicitly cast to each other (such as different numeric types). For this reason, it is generally more robust to write your queries to select specific columns. If you use SELECT *, you must modify your queries if a new column is added to one table but not to the others.

For example:

SELECT col1, col2, col3 FROM tbl1
SELECT col1, col2, col3 FROM tbl2

With table-level UNION ALL, the rows from the unioned tables are not guaranteed to process in any particular order. They may process in an interleaved fashion. If you need a particular result ordering, use ORDER BY on the outer query.


Add "EXPLAIN PLAN FOR" to the beginning of any query to get information about how it will be translated. In this case, the query will not actually be executed. Refer to the Query translation documentation for more information on the output of EXPLAIN PLAN.


For the legacy plan, be careful when interpreting EXPLAIN PLAN output, and use request logging if in doubt. Request logs show the exact native query that will be run. Alternatively, to see the native query plan, set useNativeQueryExplain to true in the query context.

Identifiers and literals

Identifiers like datasource and column names can optionally be quoted using double quotes. To escape a double quote inside an identifier, use another double quote, like "My ""very own"" identifier". All identifiers are case-sensitive and no implicit case conversions are performed.

Literal strings should be quoted with single quotes, like 'foo'. Literal strings with Unicode escapes can be written like U&'fo\00F6', where character codes in hex are prefixed by a backslash. Literal numbers can be written in forms like 100 (denoting an integer), 100.0 (denoting a floating point value), or 1.0e5 (scientific notation). Literal timestamps can be written like TIMESTAMP '2000-01-01 00:00:00'. Literal intervals, used for time arithmetic, can be written like INTERVAL '1' HOUR, INTERVAL '1 02:03' DAY TO MINUTE, INTERVAL '1-2' YEAR TO MONTH, and so on.

Dynamic parameters

Druid SQL supports dynamic parameters using question mark (?) syntax, where parameters are bound to ? placeholders at execution time. To use dynamic parameters, replace any literal in the query with a ? character and provide a corresponding parameter value when you execute the query. Parameters are bound to the placeholders in the order in which they are passed. Parameters are supported in both the HTTP POST and JDBC APIs.

In certain cases, using dynamic parameters in expressions can cause type inference issues which cause your query to fail, for example:

SELECT * FROM WHERE dim1 like CONCAT('%', ?, '%')

To solve this issue, explicitly provide the type of the dynamic parameter using the CAST keyword. Consider the fix for the preceding example: