Columnar Compression: Dictionary & Run-Length Encoding

Every analytical data warehouse — Snowflake, BigQuery, Redshift, or a raw Parquet file sitting in a data lake — leans on the same two tricks to shrink a column before it ever touches disk: dictionary encoding, which swaps repeated strings for small integer codes, and run-length encoding, which collapses a stretch of identical values into one (value, count) pair. This simulator renders a live sample of a data-warehouse column as a row of coloured 3D blocks and lets you tune row count, cardinality and how clustered/sorted the values are, watching the blocks fuse into runs or shrink into dictionary codes while the raw size, encoded size and compression ratio update from the same formulas real columnar engines use.