v.21.9Improvement
Apply LIMIT on Shards with Distributed Push Down Limit for Queries
ApplyLIMITon the shards for queries likeSELECT * FROM dist ORDER BY key LIMIT 10w/distributed_push_down_limit=1. Avoid runningDistinct/LIMIT BYsteps for queries likeSELECT DISTINCT shading_key FROM dist ORDER BY key. Nowdistributed_push_down_limitis respected byoptimize_distributed_group_by_sharding_keyoptimization. #26466 (Azat Khuzhin).
Why it matters
This feature improves efficiency for distributed queries, particularly those usingORDER BY with LIMIT and queries with DISTINCT or LIMIT BY. By pushing down LIMIT to individual shards and respecting this limit in the optimize_distributed_group_by_sharding_key optimization, it reduces unnecessary processing and data transfer, leading to faster execution times and lower resource consumption.How to use it
Enable the feature by setting the settingdistributed_push_down_limit=1 in your query or session. This will allow the LIMIT clause to be pushed down and applied on each shard during distributed query execution. No additional changes to queries are necessary.