Apache Parquet

Apache Parquet is a columnar file format for analytical workloads. Data is organized by column with per-column compression and encoding, and files carry statistics (min/max, null counts) that engines use to skip data at read time.

Parquet is the at-rest format beneath most of the ecosystem: open table formats manage Parquet files, warehouses import and export it, and every mainstream engine — Spark, Trino, DuckDB, Flink — reads it natively.

It is the default choice for raw zones and lake data; row-oriented formats like Avro complement it for records in motion, and ORC remains a comparable columnar alternative in Hive-lineage stacks.