The development team wants a safe sandbox for risky experiments: a copy of the 100 TB production table catalog.sales.orders that they can transform, rewrite, and even corrupt without consequences. Storage costs must not double, and any commits made on the copy must never affect the production table. Which approach fits?

1. A. CREATE TABLE dev.orders AS SHALLOW CLONE catalog.sales.orders — the clone reuses the existing data files while keeping independent metadata and history ✓ A
2. B. CREATE TABLE dev.orders AS DEEP CLONE catalog.sales.orders — a deep clone shares the underlying data files with the source table 💬 B
3. C. CREATE TABLE dev.orders AS SELECT * FROM catalog.sales.orders — a full physical copy is unavoidable whenever dev writes must remain isolated from production 💬 C
4. D. A shallow clone works only for read-only experiments: it keeps inheriting every later commit made on the source, so dev writes would flow back 💬 D

Reveal answer

Correct answer: CREATE TABLE dev.orders AS SHALLOW CLONE catalog.sales.orders — the clone reuses the existing data files while keeping independent metadata and history

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