What is Columnar Storage

Definition

Columnar storage is a data layout that stores values from each column together rather than storing complete rows sequentially, so analytical queries can read only the columns they need and benefit from strong compression on similar adjacent values.

« Back to Glossary Index

Reads only the columns a query needs, slashing I/O for analytical workloads
Compresses far better because adjacent values in a column are similar in type and range
Speeds aggregations and scans common in reporting and BI
Pairs with vectorised execution engines for high analytical throughput

Real World Example

A data warehouse stores sales facts in columnar format, so a query summing revenue reads only the revenue column across billions of rows instead of loading every field of every row, returning results far faster.

FAQs

Why is columnar storage fast for analytics?

Analytical queries usually touch a few columns, and columnar layout lets the engine read just those columns rather than entire rows.

Why does columnar storage compress well?

Values within a single column share data type and similar ranges, which compression algorithms exploit far more effectively than mixed row data.

When is row storage better than columnar?

For transactional workloads that read and write whole records at once, row storage is more efficient.