Explain the difference between a materialized dataset and a virtual dataset in Foundry.

• A. Materialized datasets store computed data in memory only.
• B. Materialized datasets duplicate the data in multiple files.
• C. Materialized datasets are virtual and have no persistence.
• D. Materialized dataset stores computed data persistently; a virtual dataset is computed on demand or via a query layer without persisting the results.

Answer: (D)

The key idea is whether the results of the transformation are stored or computed each time. A materialized dataset writes the computed results to storage, so you get a persistent, query-fast snapshot you can refresh on schedule or on demand. It’s like taking a snapshot of the transformed data and keeping it in a fixed place.

A virtual dataset, on the other hand, defines how to derive data but doesn’t store the results. When you query it, the system computes the results on the fly (often by running a query layer against the underlying sources or other datasets). This keeps data up-to-date because there’s no separate stored copy, but it can be slower to query since it’s computed at request time.

So the best description is that a materialized dataset stores computed data persistently, while a virtual dataset is computed on demand without persisting the results.