A company has packaged a stateless HTTP service as a container. It needs managed scheduling, automatic replacement of failed tasks, and scaling across two Availability Zones. The team does not want to patch or size EC2 container hosts. Which design best meets these requirements?

1. Use an ECS service on Fargate across two Availability Zones behind an ALB, with service auto scaling.
2. Use ECS on an EC2 capacity provider and assume EC2 host patching is no longer the team's responsibility.
3. Run the container on one EC2 instance with a local restart script.
4. Run the container only as a manually started standalone task with no service definition.