Snapshots are one of those infrastructure features people only think about right after they needed one.
On Excloud, a snapshot is a point-in-time copy of a block volume. That makes it useful before a risky migration, before a package upgrade, before changing a filesystem, or before turning one good machine image into several new machines.
What snapshots protect you from
Snapshots help with operational mistakes.
Good examples:
- A deploy writes bad data to disk.
- A package upgrade breaks a service.
- A migration needs rollback.
- A config change makes a VM unbootable.
- You want to clone a known-good volume into a new VM.
The workflow is intentionally plain:
exc compute snapshot create --volume_id <volume_id>After that, restore from the snapshot when you need a new volume or a new root volume source.
What snapshots do not protect you from
Snapshots are not a complete backup strategy by themselves.
They do not automatically prove that your application can recover. They do not replace database-aware backups. They do not give you a tested restore runbook. They do not help if nobody knows which snapshot matters.
For databases, a crash-consistent volume snapshot may still need database recovery on boot. For critical data, pair snapshots with logical backups, export paths, and restore tests.
Backups vs snapshots
A snapshot is a fast infrastructure primitive. A backup is an operational promise.
A backup strategy answers:
- What is copied?
- How often?
- Where is it stored?
- Who can delete it?
- How long is it retained?
- How often is restore tested?
Snapshots can be part of that strategy, but they are not the whole strategy.
A useful snapshot habit
Before a risky operation, take a snapshot and name the reason in your deployment notes:
exc compute snapshot create --volume_id <volume_id>Then write down:
- The source VM.
- The volume ID.
- The snapshot ID.
- The reason.
- The expected cleanup date.
The cleanup date matters. Snapshots that nobody understands become expensive clutter.
The mental model
Use snapshots as a rollback handle, a clone source, and a safety point before risky work. Use backups as the durable recovery system for data you cannot afford to lose.