Redshift's explanation is pretty good. Among other things, if you only need a few columns there are entire blocks of data you don't need to touch at all. https:

Redshift's explanation is pretty good. Among other things, if you only need a few columns there are entire blocks of data you don't need to touch at all. https://docs.aws.amazon.com/redshift/latest/dg/c_columnar_storage_disk_mem_mgmnt.html https://docs.aws.amazon.com/redshift/latest/dg/c_columnar_st... It's truly magical when you scope down a SELECT to the columns you need and see a query go blazing fast. Or maybe I'm easily impressed.