Glossary

Columnar storage

A file and database layout that stores each column's values together, instead of storing each row together.

Columnar storage organizes data by column rather than by row: all values for one field are stored contiguously on disk, in contrast to row-oriented storage, where an entire record is stored together. Analytical formats such as Apache Parquet and most cloud data warehouses use columnar storage internally.

The benefit shows up in analytical queries, which typically read a handful of columns out of a table with dozens — a columnar layout lets the engine skip every column not referenced in the query entirely, and because values within one column tend to be similar, they compress far better than mixed row data. Row-oriented storage remains better for transactional workloads that read or write whole records at once, such as an application fetching a single customer's full profile.

Columnar storage is what makes scanning billions of rows for an aggregate query practical without an index, and it is foundational to massively parallel processing warehouses and modern query engines. A common misunderstanding is treating columnar formats as universally faster; for point lookups or frequent single-row updates, row-oriented storage is still the better fit, which is why operational databases remain row-based even as analytics has moved to columnar.

Last reviewed September 22, 2026

In the index now

Related terms

Apache Parquet An open-source, columnar file format for storing large analytical datasets efficiently in disk or object storage. Apache Iceberg An open table format adding ACID transactions, schema evolution, and time travel to files stored in a data lake. Query engine The software component that parses, plans, and executes queries against one or more data sources. Massively parallel processing An architecture that splits a query across many independent compute nodes, each processing its own data slice. Data lake A storage repository that holds raw data of any format cheaply, with structure applied only when the data is read. Partitioning Physically splitting a table's data into segments, usually by a column like date, so queries can skip irrelevant ones.

Related tools

Apache Druid Open-source real-time analytics database that ingests streaming and batch data for interactive dashboards. ClickHouse Open-source columnar database built for sub-second aggregate queries over billions of rows. Exasol In-memory, massively parallel analytics database available as a managed cloud service or self-hosted software. SingleStore Distributed SQL database combining row and columnar storage for hybrid transactional/analytical (HTAP) workloads. OpenText Vertica Columnar MPP analytics database, now owned by OpenText, deployable on-premises, in Kubernetes, or across major clouds.

Related guides

How to choose a cloud data warehouse Start from your cloud, your pricing tolerance and your workload shape — general BI, high-concurrency apps, or enterprise mixed workloads — not the feature list. How to choose a real-time OLAP database Real-time OLAP databases trade operational complexity for speed on fresh data — the decisions are architecture, query concurrency, and who runs it. How to reduce data warehouse costs Where usage-based warehouse spend actually goes, and the levers — query, storage, scheduling — that reliably bring it down.