First, sorry for the late reply Arli!

Okay, I agree about the column-oriented storage but the RowGroups and Stats are so key for the Parquet format that without explaining them we are missing 2 out of 3 reasons why Parquet even exists :)

The third is the implementation of the Dremel paper but this is not so key for the initial grasp of the Parquet format and can be explained separately because one needs to explain definition and repetition levels which is far from trivial...