Join node

The join node is a performance- and memory-optimized implementation of full-, inner-, left-, semi-, and anti-joins.

The join appears in the query plan as Parallel S64 DA Hash Join with either two shuffles below the join node or one S64 DA Duplicator below the join node, depending on the data size. It also has some extra information listed separately such as Join Type and Inner Unique. When you run the query with EXPLAIN (ANALYZE, VERBOSE), the output includes extra statistics about how the join ran, such as memory consumption and the number of rows processed.

The join is only supported when shuffling is also supported and enabled.


swarm64da.enable_join (boolean)

Enables or disables the join. The default is on.