Sort/Cluster/Distributed By
This documentation is for an unreleased version of Apache Flink. We recommend you use the latest stable version.

Sort/Cluster/Distributed by Clause #

Sort By #

Description #

Unlike ORDER BY which guarantees a total order of output, SORT BY only guarantees the result rows with each partition is in the user specified order. So when there’s more than one partition, SORT BY may return result that’s partially ordered.

Syntax #

query: SELECT expression [ , ... ] FROM src sortBy
sortBy: SORT BY expression colOrder [ , ... ] 
colOrder: ( ASC | DESC )

Parameters #

  • colOrder

    it’s used specified the order of returned rows. The default order is ASC.

Examples #

SELECT x, y FROM t SORT BY x;
SELECT x, y FROM t SORT BY abs(y) DESC;

Distribute By #

Description #

The DISTRIBUTE BY clause is used to repartition the data. The data with same value evaluated by the specified expression will be in same partition.

Syntax #

distributeBy: DISTRIBUTE BY expression [ , ... ]
query: SELECT expression [ , ... ] FROM src distributeBy

Examples #

-- only use DISTRIBUTE BY clause
SELECT x, y FROM t DISTRIBUTE BY x;
SELECT x, y FROM t DISTRIBUTE BY abs(y);

-- use both DISTRIBUTE BY and SORT BY clause
SELECT x, y FROM t DISTRIBUTE BY x SORT BY y DESC;

Cluster By #

Description #

CLUSTER BY is a short-cut for both DISTRIBUTE BY and SORT BY. The CLUSTER BY is used to first repartition the data based on the input expressions and sort the data with each partition. Also, this clause only guarantees the data is sorted within each partition.

Syntax #

clusterBy: CLUSTER BY expression [ , ... ]
query: SELECT expression [ , ... ] FROM src clusterBy

Examples #

SELECT x, y FROM t CLUSTER BY x;
SELECT x, y FROM t CLUSTER BY abs(y);