Glossary

Data lake

A storage repository that holds raw data of any format cheaply, with structure applied only when the data is read.

Also called: DL, data lake storage

A data lake is a storage layer, usually built on low-cost object storage, that holds raw data in its native format: structured tables, semi-structured logs and JSON, and unstructured files like images or text. Nothing is transformed on the way in, which keeps ingestion cheap and fast even at very large scale.

The defining difference from a data warehouse is when structure is applied. A warehouse enforces schema on write, through ETL; a lake defers that decision, applying schema on read so the same raw data can support multiple downstream uses, from ad hoc data science to later ELT into a modeled warehouse. This flexibility is also the main risk: without discipline, a lake accumulates data nobody can find or trust.

Lakes matter for organizations that need to retain large volumes of raw or unstructured data cheaply, or that want to keep options open before committing to a schema. The common pitfall is the "data swamp": data landing in the lake with no data catalog, no ownership and no lineage, so it becomes unusable. Pairing a lake with cataloging and governance practices, or moving to a data lakehouse, addresses this.

Last reviewed September 19, 2026

In the index now

Related terms

Data warehouse A centralized, structured database optimized for fast historical reporting rather than day-to-day transactions. Data lakehouse A storage architecture that adds warehouse-style structure and transactions directly on top of low-cost data lake storage. Data pipeline The automated sequence of steps that moves data from source systems to a destination, transforming it along the way. Data catalog A searchable inventory of an organization's data assets, with descriptions, ownership and usage attached. ETL (extract, transform, load) A data integration pattern that transforms data before loading it, unlike ELT which transforms after loading. ELT (extract, load, transform) A data integration pattern that loads raw data first and transforms it afterward using the warehouse's own compute.

Related tools

Apache Hudi Open-source table format optimized for high-volume upserts, deletes, and incremental data-lake pipelines. Apache Iceberg Open-source table format that adds warehouse-style transactions, schema evolution, and time travel to data-lake files. AWS Data Exchange AWS-native marketplace for finding, subscribing to and billing third-party datasets delivered through S3, Redshift or APIs. Azure Event Hubs Microsoft's managed big-data streaming service, exposing a Kafka-compatible endpoint for event ingestion on Azure. Azure Machine Learning Microsoft's cloud platform for building, training, and operationalizing machine learning and generative AI models. Azure Synapse Analytics Microsoft's unified analytics service combining a dedicated SQL data warehouse, serverless SQL, and Apache Spark pools.

Related guides

How to choose a lakehouse platform Choosing a lakehouse means two decisions, not one — which open table format your data sits in, and which platform manages and queries it. How to choose a log analytics tool Pick a log platform by who owns it, whether you index everything or only labels, and how the pricing model behaves as volume grows. How to choose a SIEM Choose a SIEM by deployment model, whether detection is rule-based, behavioral, or code, and how the storage architecture handles cost as retention grows.