Snapshots (instance vs volume)
Coming from another cloud?
▸AWS·EBS Snapshots
EBS Snapshots
- Per-EBS volume, incremental in S3; not full instance image.
- Via EC2 CreateSnapshot API, not Nova/Cinder.
- Async pending state, volume usable during snapshot.
- Billed per GB-month stored.
▸DigitalOcean·Volume Snapshots
Volume Snapshots
- Proprietary API /v2/volumes/{volume_id}/snapshots instead of OpenStack /v3/{project_id}/snapshots.
- Explicitly crash-consistent only, no automatic filesystem consistency (requires manual power-off/sync); OpenStack snapshots can use quiescing drivers for apps like databases.
- Snapshots billed separately based on block-level used size (may exceed filesystem due to trim); OpenStack typically similar. See the DigitalOcean pricing page for current rates.
- Can create new volumes only in same region as snapshot; OpenStack snapshots more flexible for cross-region with upload/download.
▸Google Cloud·Standard/Archive Snapshots
Standard/Archive Snapshots
- Crash-consistent point-in-time copies stored incrementally in Cloud Storage (geo-redundant by default), unlike Cinder snapshots which depend on backend (often full copies).
- Two types: Standard (faster recovery) and Archive (cheaper long-term); supports snapshot schedules (standard only).
- Global by default, shareable across projects/regions/zones with network egress costs for cross-region restore; regional scoping (preview) for location control.
- Frequency limit: max 6 snapshots per disk every 60 minutes.
Snapshots (instance vs volume)
Snapshot means different things depending on which service creates it. On Quake AI, separate the two cases before you plan backup or migration.
Instance snapshots#
An instance snapshot captures the instance root disk (and related metadata) into a bootable image. Use it to clone a configured server, roll back after a risky change, or build a golden image pipeline.
Nova and Glance handle this path. The result is an image you can launch as a new instance.
Volume snapshots#
A volume snapshot is a point-in-time copy of a block volume managed by Cinder. Use it for database disks, attached data volumes, and backup schedules that target storage independently of the instance lifecycle.
Volume snapshots can be restored to new volumes or used as clone sources. They do not automatically replace instance images unless you design that workflow.
Provider vocabulary#
| Provider phrase | Usually means on Quake AI |
|---|---|
| AWS EBS snapshot | Volume snapshot |
| AWS AMI from instance | Instance snapshot → image |
| GCP disk snapshot | Volume snapshot |
| DigitalOcean volume snapshot | Volume snapshot |
| "Snapshot the server" (ambiguous) | Clarify: instance image vs attached volume |
What to read next#
- Instance snapshots: capture and restore instance state
- Volumes: volume snapshots and clones
- Backup and restore an instance: operational backup workflow