Reconsider which aggregations can support parallel execution #98669
Labels
:Analytics/Aggregations
Aggregations
>enhancement
Team:Analytics
Meta label for analytical engine team (ESQL/Aggs/Geo)
Currently parallel execution isn't enabled by default, but even when it is enabled then the following aggregations are set up the never execute in parallel:
scripted_metric
,parent
,terms
,multi_terms
, andsignificant_terms
.There two main raisons for this is either precision of an aggregation will be different when execution in parallel mode or the aggregation couldn't be accessed from different threads. For latter, as part of #98204, the post collection happens on the same thread as the actual collection.
For the former reason, we properly need to get a better feeling what the impact in precision is in practise for the aggregations in question (for
terms
,multi_terms
andsignificant_terms
) and update the documentation about the additional loss in precision (next to the loss of precision when running these aggs on multiple shards).The text was updated successfully, but these errors were encountered: