Phase Verbs / Action Phrases
| Phrase | Meaning |
|---|---|
| hibernate (an instance) | freeze the OS and save RAM state to disk |
| dump (RAM to EBS) | write in-memory data to the root EBS volume |
| preserve (in-memory state) | keep the running processes/data intact across a stop |
| warm (a cache) | pre-load data into memory at startup; skipped when hibernating |
| freeze / resume (the OS) | pause execution and restore it exactly as it was |
Technical Vocabulary
| Term | Definition |
|---|---|
| Hibernate | EC2 state where RAM is written to the encrypted root EBS volume; instance resumes from that state |
| RAM dump | A snapshot of all data currently in memory, written to disk |
| Root EBS volume | The EBS volume that holds the OS; used as RAM dump target during hibernate |
| In-memory state | All processes, loaded libraries, and cached data currently running in RAM |
| Stop (EC2) | Shut down instance; EBS data persists; RAM lost; cold boot on next start |
| Terminate (EC2) | Delete instance; EBS root volume deleted by default; any additional volumes may persist |
| Hibernate (EC2) | Pause instance; RAM written to EBS; resumes without OS reboot |
Key Concepts
EC2 Instance States Comparison
| Action | OS Reboot | RAM | EBS Root | Boot Time |
|---|---|---|---|---|
| Stop → Start | Yes, full OS boot | Lost | Kept | Slow (full boot + app init) |
| Terminate | N/A | Lost | Deleted (by default) | N/A |
| Hibernate → Start | No (OS is frozen) | Restored from EBS | Kept | Fast (RAM restored, no reinit) |
How Hibernate Works (Step by Step)
- Instance is running with data in RAM
- Hibernate command issued → instance enters “stopping” state
- RAM is dumped (written) to the root EBS volume
- Instance is shut down; RAM disappears
- Root EBS volume still holds the RAM dump
- On next start: RAM is loaded from EBS back into memory
- Instance resumes exactly where it left off — as if it never stopped
Requirements for Hibernate
- Root volume must be EBS (not Instance Store)
- Root EBS volume must be encrypted
- Root EBS volume must be large enough to hold the entire RAM
- Instance RAM size: < 150 GB
- Not supported on bare metal instances
- Supported OS: Linux, Windows
- Available for: On-Demand, Reserved, Spot instances
- Max hibernation time: 60 days
Use Cases
- Long-running processes that must not be interrupted
- Services with long initialization — skip re-initialization on restart
- Fast boot requirement — hibernate and resume instead of cold start
- Saving RAM state for later resume
Exam Tips
- Hibernate = RAM saved to encrypted EBS → fast resume (OS never restarts)
- Root EBS volume must be encrypted + large enough for RAM
- Max hibernate duration: 60 days
- Not available for bare metal instances
- Key difference from Stop: Stop = cold boot; Hibernate = warm resume
- Use hibernate when apps take a long time to initialize and you want to avoid that overhead