Skip to content

Column Mapping

Column Mapping allows mapping physical columns names to their logical ones for reading and writing parquet data files.

Column Mapping is enabled for a delta table using delta.columnMapping.mode table property.

Column Mapping turns ALTER TABLE RENAME COLUMN and ALTER TABLE CHANGE COLUMN commands into logical changes at metadata level (leading to no changes to the physical column names in parquet files and a mere transactional metadata update in a transaction log).

Column Mapping uses the metadata of a StructField (Spark SQL) to store the logical column name under the delta.columnMapping.physicalName metadata key.

Column Mapping is available as of Delta Lake 1.2.0.

Minimum Required Protocol

Column Mapping requires the minimum protocol version to be 2 for readers and 5 for writers.


Demo: Column Mapping

Learn More

  1. Delta Lake 1.2 - More Speed, Efficiency and Extensibility Than Ever