v.19.15Experimental Feature

Implement (in Memory) Merge Join Variant That Does Not Change Current Pipeline

Official CH changelog text
Implement (in memory) Merge Join variant that does not change current pipeline. Result is partially sorted by merge key. Set partial_merge_join = 1 to use this feature. The Merge Join is still in development. #6940 (Artem Zuikov)
Implemented an in-memory variant of the Merge Join that integrates without altering the current query processing pipeline and produces results partially sorted by the merge key.

Why it matters

This feature introduces a new Merge Join algorithm variant that operates in memory, improving join performance while maintaining compatibility with existing query pipelines. It helps users efficiently perform join operations with results partially sorted on the join key, facilitating downstream processing.

How to use it

Enable the feature by setting the setting partial_merge_join to 1. This can be done by adding the following to your query or configuration:

SET partial_merge_join = 1


Once enabled, Merge Join will execute using this new in-memory variant.

Related resources

PreviousNext