A documented pilot-light recovery environment already replicates critical data to a secondary Region. Its application infrastructure is defined but most compute is not running. During a primary-Region outage, which sequence best follows this recovery model?

1. Provision or start the application compute, validate the replicated data and service health, then redirect traffic according to the runbook.
2. Redirect all traffic immediately and assume the inactive application compute can serve requests.
3. Treat the recovery Region as a full-capacity active-active environment without making any changes.
4. Delete the replicated data and wait for the primary Region to become available before starting recovery.