How Select query works?

as per my reading Druid do a lot of jobs about aggregating data on maps and generating segments that contains these aggregated and compressed data.

I was than surprised on reading that there was a way with Select query to have back raw data.

Does it means that Druid also store raws data? Is there a way to disable raws in order to reduce the used space?



Druid summarizes the raw data being ingested and does rollup based on the timestamp(truncated by rollupGranularity) and dimension values.
When you do select query you will get rolled up data instead of raw data.

druid docs has more info on rollup -