What does an ORM do underneath?

Definition:

An ORM translates object-oriented data operations into database commands and maps results back into application objects. It reduces repetitive persistence code, but the database still executes real SQL with real locks, indexes and query costs. Understanding the generated query is essential when debugging correctness or performance.

Key points:

• It generates SQL, sends it through a driver and maps returned rows to objects.

• Inspect generated SQL to catch N+1 queries, large joins and missing filters.

• Use raw SQL when it is clearer or gives needed control; it is not automatically faster.

• Lazy relation access can issue extra SQL long after the original query, so inspect query logs around the whole request.

Interview point:

• Use the ORM for convenience while retaining the ability to reason about plans, transactions and raw SQL when needed.