Basic Information
| Field | Details |
|---|---|
| English name | Amazon EBS |
| Full name | Amazon Elastic Block Store |
| Chinese description | 弹性块存储 |
| Japanese description | Amazon EBS（EC2 向けの永続ブロックストレージ） |
| Exam frequency | ⭐⭐⭐⭐⭐ |
| Often confused with | Instance Store / EFS / S3 |
In one sentence
EBS behaves like a cloud disk attached to EC2 in the same Availability Zone. It fits operating systems, databases, and low-latency random I/O.
Core purpose
- An EBS volume is an AZ-scoped resource; moving across AZs normally uses a snapshot and a new target volume.
- Stopping EC2 preserves the volume; termination behavior depends on Delete on Termination.
- IOPS measures operations per second, while throughput measures data transferred per second.
- A volume belongs to its AWS account, Region, and AZ, not privately to the IAM user who created it.
- Visibility depends on the current account, Region,
ec2:DescribeVolumespermission, console filters, and resource state. - Creating a volume successfully does not guarantee permission for later Describe, Modify, or Delete actions.
Exam focus
- Use Instance Store for temporary, reproducible data and EBS for persistent system or database storage.
- gp3/gp2 are general purpose, io2/io1 target critical high-IOPS workloads, and st1/sc1 favor sequential throughput or cold data.
- Snapshots are incremental block-level point-in-time backups; DLM automates their lifecycle.
Common misconceptions
- Persistent does not guarantee that a volume survives instance termination.
- EBS is not a naturally multi-AZ shared file system.
- Console multi-session does not change resource ownership or merge permissions.
Key takeaway
Instance Store for temporary local data; EBS for persistent disks; snapshots rebuild across AZs; the account owns the volume and policy controls access.
Related services
EC2, EBS Snapshot, DLM, AWS Backup, KMS.