Parquet Encoding
Parquet encoding is the set of encoding schemes used by the Apache Parquet columnar format, including Plain, Dictionary, RLE/Bit-Packed, and Delta Binary Packed.
Purpose
It selects the optimal encoding per column data type in columnar storage to maximize compression and scan speed.
Mechanism
Each column chunk selects an encoding based on its statistics: repeated values use Dictionary, low-cardinality data uses RLE/Bit-Packed, incrementing integers use Delta Binary Packed, and floats use Byte-Array Split.
Use cases
- Parquet columnar files
- Hadoop/Spark data lakes
- Analytical data storage