To create a table in Microsoft Fabric Warehouse that references data from the Lakehouse without duplicating the data, the best approach is to use cross-database queries or external tables/views. These methods allow the Warehouse to query Lakehouse data directly without physically copying it.
Options to Create Tables Without Data Duplication:
1. Use Cross-Database Queries (No Physical Copy)
Fabric workspace supports querying Lakehouse tables directly in Warehouse SQL by using a three-part naming convention:
SELECT * FROM lakehouse_database.schema.table
This allows you to access live Lakehouse data from within the Warehouse query environment without copying data.
You can create views in the Warehouse that reference Lakehouse tables, essentially creating virtual tables with zero data duplication:
CREATE VIEW warehouse_schema.view_name AS SELECT * FROM lakehouse_database.schema.table;
2. Create External Tables (Virtual References)
External tables in Fabric can reference Lakehouse storage paths without importing data to the Warehouse.
This functionality exposes the Lakehouse data files as tables accessible from within the Warehouse without physically copying them.
You define the external table pointing to the Delta Lake storage location of the Lakehouse.
3. Cloning Tables (Zero-Copy Clones)
Fabric supports zero-copy cloning of tables where a cloned table shares the same underlying storage and doesn’t duplicate data.
This is helpful for test or development environments to create independent but storage-efficient table clones:
CREATE TABLE warehouse_schema.clone_table AS CLONE lakehouse_database.schema.original_table;
When to Use Each:
Use cross-database queries and views when you want live access to Lakehouse data inside Warehouse queries without replication.
Use external tables if you want a table in Warehouse referencing Lakehouse storage for broader integration.
Use clones when you want independent tables with minimal storage without affecting the original data.
