S3 select parquet gzip vs json GZIp

Question

I am doing some analysis for S3 query performance for log data on S3 storage. In theory parquet should be more performant if I query only some columns not whole