data lake
/DAY-tuh layk/ · noun · Development · Origin: 2010
Definitions
A centralized storage repository that holds vast amounts of raw data in its native format — structured, semi-structured, and unstructured — until it is needed for analysis. Unlike a data warehouse, a data lake stores data without requiring a predefined schema. If not properly governed, data lakes can devolve into 'data swamps' where data becomes unusable.
In plain English: A massive storage pool where you dump all your raw data as-is — like throwing everything into a giant filing cabinet now and organizing it later when you need it.
Example: They dumped all their raw event logs, PDFs, and IoT sensor data into the data lake on S3, then ran Spark jobs to extract insights as needed.
Origin Story
The storage strategy that says 'dump everything, ask questions later'
Pentaho CTO James Dixon coined **data lake** in a 2010 blog post, contrasting it with data warehouses. A warehouse is like bottled water -- cleansed, filtered, and packaged. A data lake is the natural body of water: raw, unprocessed, and potentially vast.
The idea was appealing: store everything in raw form (usually on Hadoop/HDFS) and let data scientists explore it later with their own tools. No upfront schema design needed. Companies dumped logs, CSVs, JSON, images, and videos into the same lake.
The reality was messier. Without governance, data lakes became **data swamps** -- vast repositories of undocumented, untrusted data nobody could find or understand. The industry responded with the **data lakehouse** pattern (Databricks, 2020), combining the flexibility of lakes with the structure of warehouses.
Coined by: James Dixon (Pentaho CTO)
Context: Blog post, October 2010
Fun fact: The 'data swamp' counter-metaphor became so common that it arguably overshadowed the original term. Dixon later clarified that a data lake needs proper management, but by then the industry had already learned the hard way.