lakehouse

Data Lakehouse

A storage architecture that puts warehouse-style structure and querying on top of lake-style cheap, flexible storage.

When you'd see it: Databricks vendor pitches, Snowflake's response to them, and every "modern data platform" architecture deck from the last few years. Open table formats — Delta, Iceberg, Hudi — are what make the pattern work technically.

Why it matters: Lakehouse is the industry's attempt to stop making teams choose between the lake's flexibility and the warehouse's reliability. One storage layer, two access patterns: data engineers get raw flexibility, analysts get queryable tables. Whether it delivers on that promise depends heavily on which vendor's version you're using.

Common mistakes: Reading "lakehouse" as a settled standard. It's a category the industry is still defining. Databricks and Snowflake mean meaningfully different things by the word, and the underlying table formats compete. Treat the term as a direction, not a fixed thing.

Related SQL, ETL