Do Merging of Sorted Blocks on Initiator with Distributed Group By No Merge

Why it matters

This feature improves query efficiency by reducing unnecessary merging operations on remote nodes during distributed GROUP BY queries. By limiting the merge of sorted data blocks to only the initiator server, it reduces network overhead and accelerates query execution in distributed environments.

How to use it

Enable the setting distributed_group_by_no_merge to activate this optimized merging behavior. This can be done by adding SET distributed_group_by_no_merge = 1 in your query or configuring it in the server settings.