Fabric Lakehouse vs Fabric Data Warehouse
People think a Fabric lakehouse and a Fabric warehouse are the same SQL store, and they are not. Both keep Delta tables in OneLake. A lakehouse is filled by Spark and holds loose files too. Its SQL endpoint can only read. A warehouse is built with T-SQL, which can add, change and delete rows. Pick the lakehouse for Spark teams and the warehouse for SQL teams.
Fabric Lakehouse
A Fabric store for tables and files. Spark writes to it, and SQL reads it.
A Fabric lakehouse is a store in Microsoft Fabric that holds both tables and loose files. Its Tables folder holds Delta tables, and its Files folder holds the rest. Engineers fill it with Spark notebooks and pipelines. Each one comes with a SQL endpoint, so analysts can query its tables with T-SQL. That endpoint can read the data but not change it. All of it is kept in OneLake.
Fabric Data Warehouse
The SQL warehouse in Microsoft Fabric. You load and change it with T-SQL.
Fabric Data Warehouse is the SQL warehouse inside Microsoft Fabric. You build and load it with T-SQL, the kind of SQL that SQL Server uses. Unlike a lakehouse's SQL endpoint, it can add, change and delete rows. A change that spans many tables either all happens or none of it does. Its tables are kept as Delta tables in OneLake. So Spark and Power BI can read them too.
In one line
- Fabric Lakehouse
- A Fabric store for tables and files. Spark writes to it, and SQL reads it.
- Fabric Data Warehouse
- The SQL warehouse in Microsoft Fabric. You load and change it with T-SQL.