A Linux application runs on EC2 instances across two Availability Zones. All instances need concurrent NFS access to the same durable files, including when instances are replaced. The team wants managed regional storage rather than operating file servers. Which design best meets these requirements?

1. Store the files on each instance's local instance store and copy them manually after a failure.
2. Use EFS One Zone storage in one Availability Zone and assume it provides the same multi-AZ storage availability as Regional EFS.
3. Attach one ordinary EBS volume simultaneously to all instances across both Availability Zones.
4. Use an EFS Regional file system with mount targets in the application Availability Zones and appropriate NFS security-group access.