-
Notifications
You must be signed in to change notification settings - Fork 1.1k
Description
Is your feature request related to a problem or challenge? Please describe what you are trying to do.
Currently the parquet dictionary page size limit applies to all columns of the parquet file. this is quite inflexible because we know some columns needs much larger page size than other columns, e.g., string columns should have larger dictionary page size limit than int columns.
Related to XiangpengHao/liquid-cache#227
Describe the solution you'd like
Make the configuration per column not per file.
In general, I'd like to make Parquet writer more configurable and with reasonable defaults.
My ultimate vision is to have a auto-tuner which decides the best encoding/page sizes/row group sizes for a given workload.
Eventually, such a writer should produce a parquet file that is competitive with proprietary ones.
Describe alternatives you've considered
Additional context