What is the difference between backups and snapshots?
Both snapshots and backups allow you to save data and restore it in the event of a failure or error, but they do it in different ways. The choice of approach depends on the goals: quick rollback in the same cluster, volume cloning, long-term storage or recovery in a different environment.
Quick comparison
| Criterion | Snapshots (VolumeSnapshot) | Backups (backup) |
|---|---|---|
| What is saved | Single Volume Snapshot (PVC) at a Time | Cluster/namespace/volume data and, if necessary, manifests (for example, Velero) |
| Where is it stored | Typically on the same storage system (same array, same cloud) | Often on external storage (S3, NFS, object storage) |
| Creation Speed | Usually fast (copying metadata/blocks without completely copying the data) | Depends on volume; full copying can take minutes or hours |
| Recovery | To the same cluster (new PVC from snapshot or rollback to snapshot) | To the same or another cluster; it is possible to restore an entire namespace or cluster |
| Typical Goal | Quick rollback, volume cloning, pre-operation restore point | Long-term storage, disaster recovery (DR), migration to another environment |
Snapshots (VolumeSnapshot)
Snapshot is a snapshot of a volume at a certain point in time, created by the storage system (via CSI). The snapshot data remains in the same storage infrastructure (same backend, same region/data center).
Suitable when:
- You need a restore point before a risky operation (database update, migration, mass data change).
- You need to quickly clone a volume (create a new PVC with the same data) in the same cluster.
- We need a short “done and rolled back” cycle without transferring data to external storage.
Limitations:
- The snapshot is tied to a specific storage backend; If the entire storage system fails, snapshots may not be available.
- Usually not designed for very long-term storage and storage in another location (another cluster, another region).
- Support depends on the CSI driver: oVirt CSI and vCloud Director CSI do not support snapshots. See the full list on Snapshots.
More details: Snapshots.
Backups (backup)
Backup in the context of the Shturval platform is a backup copy of cluster resources (namespaces, Kubernetes objects, volumes, etc.), which is saved to external storage (for example, S3-compatible). Commonly used backup module (for example, based on Velero), configured by the administrator.
Suitable when:
- You need long-term storage of data copies and the ability to restore them after a long time.
- It is necessary to restore a cluster or namespace after a serious failure or deletion.
- Need transfer to another environment (another cluster, another region, DR site).
- Regulatory requirements or company policies require that copies be stored outside the core infrastructure.
Limitations:
- Creating and restoring a backup usually takes longer than creating and using a snapshot.
- Requires configuration of external storage and backup policies.
Recommendations for setting up backup to restore access to clusters: Backup Recommendations.
When to use what
- Snapshots only - if a quick rollback or cloning of volumes in the same cluster is sufficient and there are no strict requirements for moving copies to external storage.
- Backups only - if you need copies in external storage, restoration of an entire cluster/namespace, or transfer to another environment.
- Snapshots and backups together - snapshots for frequent “spot” rollbacks and volume cloning, backups for regularly moving data to external storage and disaster recovery. This combined approach is often used in production.