Compare the options
| Service | Interface | Typical use |
|---|---|---|
| EBS | Block device in an AZ | OS, database disk, application volume |
| EFS | Shared NFS filesystem | Shared Linux files across instances |
| S3 | Object API | Media, backups, static assets |
| Instance store | Local temporary block storage | Re-creatable scratch/cache |
EBS is not normally a shared network filesystem. Specialized multi-attach options have limitations and require cluster-aware software. EFS requires mount targets, security groups and NFS client access; do not expose NFS to arbitrary networks. An S3 prefix looks like a folder but does not supply normal POSIX filesystem semantics.
Snapshot and AMI lab
- Create a marker file on a disposable EBS volume and flush application writes appropriately.
- Create an EBS snapshot. Record the source volume, time and encryption key.
- Restore it to a new volume in the target AZ, attach and verify the marker.
- Create an AMI from a configured lab instance. Understand the reboot/no-reboot consistency tradeoff before choosing no-reboot.
- Launch another instance from that AMI and verify the service, security groups and instance role separately. An AMI does not automatically duplicate every network or IAM setting.
Vertical scaling without losing the address
Use an Elastic IP for a stable public IPv4 when appropriate. Back up first, verify the target instance type supports the AMI architecture and required devices, stop the EBS-backed instance, change type and start. Expect downtime. An automatically assigned public address can change; the primary private IP generally persists. Confirm health and application performance after restart.
Performance
gp3 separates storage size from configurable IOPS/throughput within limits; provisioned performance still faces instance-level constraints. CPU-credit-based instances can have workload-dependent behaviour. Bigger does not guarantee faster if the bottleneck is a locked database or external API.
Verification and cleanup
Write down an actual restore test, not just a successful snapshot state. Deregistering an AMI may leave snapshots unless you intentionally remove eligible ones. Keep only recovery assets you need, and never delete a shared encryption key to save a small storage charge.
Official references
Ravindra’s Tip
AMI से server की copy बन सकती है, लेकिन networking और permissions अपने आप पूरी copy नहीं होते। नया instance launch करके verify करो।
Interview and revision check
What is the difference between an AMI and a running duplicate environment?
An AMI provides launchable image/block-device information. Network configuration, roles, secrets and dependent services still need separate setup.
Ravindra Bagale · Cloud & DevOps Academy · Handbook and project downloads