An application stores disposable session data in an ElastiCache Redis OSS replication group. It needs managed failover if the primary's Availability Zone fails. Which configuration best improves cache availability?

1. Place every primary and replica node in the same Availability Zone.
2. Use Multi-AZ with automatic failover and place at least one replica in another Availability Zone.
3. Keep only one primary node and create a daily snapshot.
4. Use DNS round robin over two unrelated writable primaries with no replication.