Comment by uwemaurer

Great! I will give it a try. I found that using DuckDB to select from the parquet files and using the Apache Arrow API to read the results is also a very fast method