Dropping columns increases query time > 10-fold


I have a fairly big time series database where each measurement has several tags. Sometimes, it’s useful to create a query that ‘ignores’ these tags, so I either use drop(columns: [“SomeTag”]) or group().

I just did a query that took around 60 seconds when dropping a column, while it took less than 5 seconds without.

Where does this performance hit come from, and is there a way to avoid this penalty?