A stateless web service runs on one EC2 instance. Each instance can handle the entire normal workload. The service must remain available if one Availability Zone fails and must replace unhealthy instances automatically. Which design provides the best improvement with minimal application changes?

1. Use an ALB and an Auto Scaling group across two Availability Zones, keeping at least one adequately sized instance in each zone.
2. Create a larger EC2 instance and take an AMI every night.
3. Run two instances in the same subnet and register both with one ALB.
4. Move the instance into a placement group with all application components in one Availability Zone.