is there a way to set up reflections on count distinct?
Instead of using approximation, can we set up aggregate refections to get the accurate count distinct?
If you use an aggregation reflection and the column is used as a dimension, you should get fast accurate count distinct. Dremio needs to maintain all the distinct values in order to calculate all the measures, and the query planner can take advantage of this.
Fast approximate distinct is an option for measures.