Years of analytics history live as Parquet datasets in the company's cloud storage lake. Governance now requires these datasets to become Delta tables, so that the lakehouse gains ACID transactions and time travel on top of them, but the data files must stay exactly where they are — no copies, no rewrites, no new storage locations. Which approach meets the requirement?

1. A. Run CREATE TABLE AS SELECT over each dataset, copying the rows into newly written Delta files under a managed location 💬 A
2. B. Run CONVERT TO DELTA, which copies the existing Parquet files into the table's managed storage location first 💬 B
3. C. Register the datasets as plain external Parquet tables instead — Parquet already guarantees ACID transactions, so converting them adds nothing 💬 C
4. D. Run CONVERT TO DELTA on each dataset — Delta builds its transaction log over the existing files in place, without copying them ✓ D

Reveal answer

Correct answer: Run CONVERT TO DELTA on each dataset — Delta builds its transaction log over the existing files in place, without copying them

Reference: https://docs.databricks.com/aws/en/delta/