Examples
Worked examples
- Is an instance
An institutional research-IT data lake holding raw genomics FASTQ files, microscopy images, and instrument logs.
- Is an instance
An astronomical observatory's S3-based data lake ingesting raw telescope outputs prior to pipeline processing.
Counter-examples
Looks similar, but isn't
- Not an instance
A small structured database is not a data lake.
- Not an instance
A curated, schema-on-write data warehouse is the contrasting pattern, not a data lake.
Editorial commentary
A data lake is a general information-systems pattern for storing large volumes of raw data in its native format — structured, semi-structured, or unstructured — without imposing a fixed schema at write time, deferring structure until a consumer actually queries the data (‘schema-on-read’). The pattern emerged in the early 2010s, notably promoted by James Dixon at Pentaho, as a counter-position to traditional data warehousing’s schema-on-write approach. Modern data lakes are typically built on cloud object storage (S3, Azure Blob, GCS), with optional table-format layers (Apache Iceberg, Delta Lake, Hudi) adding transactional and schema-management features on top, and compute engines such as Spark or Trino querying the stored data directly.
Scope note: data-lake architecture is a large, well-established data-engineering discipline with its own extensive literature — table-format internals, partitioning strategy, and query-engine tuning are best learned from that literature directly, not from a research-standards page. This entry stays narrow to why the term shows up in a research-administration or research-data-management context at all.
Where data lakes show up in research infrastructure
In research-data contexts, data lakes are typically used as a landing zone for raw, high-volume observational data — instrument telemetry, sensor logs, sequencing output — before it is curated, quality-controlled, and structured into a research-facing product. This is a deliberately earlier, rougher stage than a trusted digital repository‘s curated holdings, and a different architectural pattern from a data hub: a hub applies its harmonised model up front (schema-on-write) to serve a specific consuming community, where a lake defers structure and stores data close to its raw upstream form — see the data hub entry’s own comparison section for the fuller distinction. A data warehouse sits at the opposite end from a lake on the same schema-on-write axis, typically for structured institutional reporting rather than raw research data.
References
- Dixon J., ‘Pentaho, Hadoop, and Data Lakes’ (Pentaho blog, 2010).
- Russom P., ‘Data Lakes: Purposes, Practices, Patterns, and Platforms’ (TDWI Best Practices Report, 2017).
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="Data lake"
vocab-term-identifier="https://casrai.org/dictionary/term/data-lake" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/data-lake",
"name": "Data lake",
"identifier": "https://casrai.org/dictionary/term/data-lake",
"description": "A storage repository that holds large volumes of structured, semi-structured, and unstructured data in their native formats, deferring schema-on-write requirements so that data can be ingested cheaply and only structured at the time of read or analysis.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/data-infrastructure#set",
"url": "https://casrai.org/dictionary/term/data-lake",
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"author": {
"@id": "https://casrai.org/#editorial-team"
},
"datePublished": "2026-05-21T02:17:58",
"dateModified": "2026-09-05T14:23:19",
"inLanguage": "en-GB",
"isAccessibleForFree": true
}
