.parquet File — Apache Parquet Columnar File

Apache Parquet is a columnar storage file format designed for efficient data processing and analytics in big data ecosystems. Unlike row-oriented formats (CSV, Avro), Parquet stores data column by column, enabling dramatically faster analytical queries that only need to read specific columns from wide tables with many fields. PARQUET files use a hierarchical structure with row groups, column chunks, and pages, supporting predicate pushdown (skipping irrelevant row groups based on column statistics) and column pruning (reading only needed columns). The format supports complex nested data structures (maps, lists, structs), multiple compression codecs per column (Snappy, Gzip, LZ4, Zstd), and dictionary encoding for columns with repeated values. Parquet has become the de facto standard for analytical data lakes, used by Apache Spark, Apache Arrow, Pandas (via pyarrow/fastparquet), Presto/Trino, DuckDB, BigQuery, Athena, and Snowflake. The format's columnar design aligns perfectly with modern CPU architectures and vectorized query engines, enabling orders-of-magnitude performance improvements over row-oriented formats for analytical workloads. Parquet files are self-describing with embedded schema metadata.