What method allows you to write queries referencing tables from different lakehouses without creating additional copies?

The correct method for writing queries that reference tables from different lakehouses without creating additional copies is through a view. A view acts as a virtual table that can encapsulate complex queries, allowing users to access data stored across different locations seamlessly. By using a view, you can create a defined structure or schema that points to the underlying tables in the lakehouses, enabling you to pull data together for analysis without the need to duplicate or physically move the data.

This approach is beneficial as it ensures data consistency and reduces redundancy in the system. Additionally, views can be easily modified to reflect changes in the underlying data structures, making them a flexible solution for querying across various lakehouses.

The other methods like shortcut, dataflow, and managed tables serve different purposes in data management and integration. Shortcuts often point to existing objects, dataflows define a process for transforming and moving data, and managed tables involve physical data storage and management rather than flexible querying across different datasets. Thus, the use of views effectively meets the need for referencing multiple table sources without unnecessary duplication.