Containers in an ECS service run across several Availability Zones and must read and write one POSIX file tree that survives task replacement. Which storage should be attached?
Show answer and explanation
Answer: A
Amazon EFS provides a managed NFS file system that containers in many Availability Zones can mount at once. ECS task definitions reference the file system, and tasks on Fargate or EC2 can read and write the same data even after a task is replaced. EBS volumes attach to one instance in one Availability Zone, host bind mounts stay on a single machine, and Fargate's ephemeral storage is removed when a task stops.
Why the other options are wrong
- B. An EBS volume is tied to one Availability Zone and generally attaches to one instance at a time. Containers in other zones cannot share it, so it does not fit a shared multi-AZ file tree.
- C. A host bind mount exposes a directory on a single container instance. Tasks scheduled on other instances cannot see that data, so it is not shared storage across the service.
- D. Ephemeral Fargate storage is scratch space that is deleted when the task stops. It cannot persist data across replacements or be shared between tasks.