Glossary

Data Lake

A data lake is a large store of raw files in cheap object storage: tables, JSON, logs, images — any shape, landed as-is, with structure applied later, on read. Its economics are the appeal: storage is cheap enough to keep everything, and nothing needs modelling up front, which suits data science and ML.

Its weakness is everything a bare pile of files doesn’t provide — no transactions, no enforced schema, no reliable history. Ungoverned, a lake drifts into a data swamp nobody trusts. The modern fix is layering an open table format over the same files, which turns the lake into a lakehouse.

Go deeper: Data Lake vs Lakehouse: What Changed and Which to Use