Also written: columnar, column store, column stores
Storing a table one column after another, so a query reads only the columns it uses and each column, holding similar values, compresses well.
Try it now
Step through three questions on the same table. The lit cells are what each one reads.
Learn it properly
Columnar files and a local warehouse · Rows against columns, the encodings that shrink a column, Parquet's row groups and the ranges that let a query skip them, and SQL over files with DuckDB. About 40 minutes.
data warehouse: Where an organization keeps the data it analyzes: columnar, built for scans over history, and queried with SQL.
dictionary encoding: Listing each distinct value of a column once and replacing the values with small codes that point into the list.
DuckDB: A columnar analytics database that runs inside your program and queries Parquet and CSV files directly with SQL.
Parquet: The columnar file format used across analytics and machine learning. It is cut into row groups that record each column's smallest and largest value.
row group: A block of rows in a Parquet file, stored column by column with each column's minimum and maximum, so a query can skip the block.
run-length encoding: Writing a repeated value once with a count of how many times it comes in a row.
Keep learning
More in plain English.
1286 terms in the glossary, each with the lesson that teaches it.
Rows against columns, the encodings that shrink a column, Parquet's row groups and the ranges that let a query skip them, and SQL over files with DuckDB.
See the lesson →
Never trained an AI? Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account. Teach it to write