Data Lake vs Data Lakehouse
People think a lakehouse is a new kind of store you buy. It is a data lake with a table format on top. The files stay in the same cheap place. The table layer adds safe updates, fixed shapes and old versions, so the lake can serve reports too.
Data Lake
A big, cheap store of raw files of every kind. Shape comes later.
A data lake is a cheap, huge store for raw data of every kind. Tables, JSON, logs, photos and video all go in as files. It usually sits on cloud object storage. Nothing is cleaned or shaped on the way in. Tools read the files later and apply a shape when they query them.
Data Lakehouse
A data lake with warehouse tools on top. One copy of the data serves all.
A data lakehouse puts warehouse features on top of a data lake. The data stays as open files, often Parquet, on cheap storage. A table format adds a log that tracks which files make up each table. That log brings safe updates, fixed schemas and time travel. So one copy of data can serve both reports and machine learning.
In one line
- Data Lake
- A big, cheap store of raw files of every kind. Shape comes later.
- Data Lakehouse
- A data lake with warehouse tools on top. One copy of the data serves all.