Skip to main content

Join Table Engine

Optional prepared data structure for usage in JOIN operations.


This is not an article about the JOIN clause itself.

Creating a Table

CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster]
name1 [type1] [DEFAULT|MATERIALIZED|ALIAS expr1] [TTL expr1],
name2 [type2] [DEFAULT|MATERIALIZED|ALIAS expr2] [TTL expr2],
) ENGINE = Join(join_strictness, join_type, k1[, k2, ...])

See the detailed description of the CREATE TABLE query.

Engine Parameters


join_strictnessJOIN strictness.


join_typeJOIN type.

Key columns

k1[, k2, ...] – Key columns from the USING clause that the JOIN operation is made with.

Enter join_strictness and join_type parameters without quotes, for example, Join(ANY, LEFT, col1). They must match the JOIN operation that the table will be used for. If the parameters do not match, ClickHouse does not throw an exception and may return incorrect data.

Specifics and Recommendations

Data Storage

Join table data is always located in the RAM. When inserting rows into a table, ClickHouse writes data blocks to the directory on the disk so that they can be restored when the server restarts.

If the server restarts incorrectly, the data block on the disk might get lost or damaged. In this case, you may need to manually delete the file with damaged data.

Selecting and Inserting Data

You can use INSERT queries to add data to the Join-engine tables. If the table was created with the ANY strictness, data for duplicate keys are ignored. With the ALL strictness, all rows are added.

Main use-cases for Join-engine tables are following:

  • Place the table to the right side in a JOIN clause.
  • Call the joinGet function, which lets you extract data from the table the same way as from a dictionary.

Deleting Data

ALTER DELETE queries for Join-engine tables are implemented as mutations. DELETE mutation reads filtered data and overwrites data of memory and disk.

Limitations and Settings

When creating a table, the following settings are applied:













Disables persistency for the Join and Set table engines.

Reduces the I/O overhead. Suitable for scenarios that pursue performance and do not require persistence.

Possible values:

  • 1 — Enabled.
  • 0 — Disabled.

Default value: 1.

The Join-engine tables can’t be used in GLOBAL JOIN operations.

The Join-engine allows to specify join_use_nulls setting in the CREATE TABLE statement. SELECT query should have the same join_use_nulls value.

Usage Examples

Creating the left-side table:

CREATE TABLE id_val(`id` UInt32, `val` UInt32) ENGINE = TinyLog;
INSERT INTO id_val VALUES (1,11)(2,12)(3,13);

Creating the right-side Join table:

CREATE TABLE id_val_join(`id` UInt32, `val` UInt8) ENGINE = Join(ANY, LEFT, id);
INSERT INTO id_val_join VALUES (1,21)(1,22)(3,23);

Joining the tables:

SELECT * FROM id_val ANY LEFT JOIN id_val_join USING (id);
│ 1 │ 11 │ 21 │
│ 2 │ 12 │ 0 │
│ 3 │ 13 │ 23 │

As an alternative, you can retrieve data from the Join table, specifying the join key value:

SELECT joinGet('id_val_join', 'val', toUInt32(1));
┌─joinGet('id_val_join', 'val', toUInt32(1))─┐
│ 21 │

Deleting a row from the Join table:

ALTER TABLE id_val_join DELETE WHERE id = 3;
│ 1 │ 21 │