A 50 TB PostgreSQL database must move to Aurora PostgreSQL, and the business allows only a few minutes of downtime at cutover. Which approach fits?
Show answer and explanation
Answer: B
DMS performs a full load while the source stays online, then keeps replicating changes captured from the source. When the target is nearly in sync, the team stops writes briefly, lets the final changes apply, and switches the application. A dump and restore would require a long outage for 50 TB, a physical copy goes stale as soon as it is taken, and copying raw data files is not a supported way to load Aurora.
Why the other options are wrong
- A. A dump and restore must run with writes stopped for the entire copy, which for 50 TB means hours of downtime rather than minutes.
- C. A physical copy is out of date as soon as it is taken, and reapplying every change manually is error-prone and far slower than continuous replication.
- D. Aurora manages its own storage layer, so raw PostgreSQL data files cannot be placed onto it, and doing so would not produce a consistent, working database.