A read replica falls behind during a write surge. The application serves stale data from the replica. Which approach best fits?

1. Bound application connection pools and assess a suitable pooling strategy.
2. Inspect query plans and supported Query Insights diagnostics.
3. Monitor replication lag and route freshness-critical reads to an appropriate consistent source.
4. Coordinate credential version rollout, client refresh, and retirement of the old credential.