-
Type: Improvement
-
Resolution: Unresolved
-
Priority: Minor - P4
-
None
-
Affects Version/s: None
-
Component/s: None
-
Query Integration
Currently a sharded continuous or discrete $percentile does all of the work in mongos. These two percentile methods are already quite slow because they need to sort all of the data and then look at all of it at the same time for a precise result, but we could speed this up by pushing down the sorting to the shards.