A disaster-recovery rehearsal must validate the recovered application without disrupting production or connecting duplicate servers to the production network. Which approach fits?

1. Fail over production without notifying its owners
2. Review only the backup bill without running recovery
3. Delete the primary before every rehearsal
4. Test failover into an isolated recovery-test network