Two kinds of recovery

A snapshot is operational recovery: a fast rollback inside the cluster after a bad migration or a wrong command, done in minutes. A backup is an encrypted copy on separate object storage outside the SDS: a local backup stays in the same region and survives losing the storage cluster, an off-site DR backup survives the region and the provider too. Most people run both a snapshot and a backup, one for the mistakes they make and one for the failures they do not control.

A backup bills the compressed archive it ships. A snapshot bills the full provisioned disk size.

Example: a 100 GB disk with 20 GB written compresses to roughly 12-15 GB in a backup archive, billed at that size. A snapshot of the same disk is billed on the full 100 GB provisioned, whatever it actually holds.

Backup

Disaster recovery.

Encrypted copy living off the cluster. Store one copy on the local ServersCamp S3, or fan out to independent providers in other countries for true off-site DR.

A local copy survives losing the whole cluster. A DR copy lives in another country, beyond the region entirely.

€35 /TB per month local S3

€70 /TB per month off-site DR backend

Billed on the compressed archive, one copy per backend. Turn on several backends and each copy is charged at its own rate.

Snapshot

Operational recovery.

Point-in-time full disk copy on the same SDS cluster, three live replicas.

Rolls back your own mistakes in minutes. Same failure domain, so not DR on its own.

€35 /TB per month provisioned disk

€0.035 per GB of the full disk size

Billed on the full provisioned disk, whatever the disk holds. Kept on the cluster until you delete it.

Verified backups optional

Turn on verification and every backup is restored into a sandbox, booted, and probed for the guest agent and SSH. A verified backup is proven bootable, with a measured restore time on record.

Backups usually get their first real restore during an incident. Verification does that restore ahead of time instead: a passing backup has already booted on our side, and its recovery time is on record before you ever reach for it.

€15 per TB of the archive it downloads, charged once per run. A 1 TB archive verified nightly costs €15 each night it runs. Verifying a 100 GB archive costs €1.50.

Storage is billed monthly, prorated by the hour: each copy is charged for its size every hour it exists, summed across the month. Delete it and billing stops at the next hour. Verify is a per-run verification charge, applied when the boot-test runs.

Both snapshots and backups restore into something new. Turn either one into a fresh disk to spin up a new server, or attach it as a cloud volume, without touching the original.

What each one survives

A copy can live in three places: a snapshot on the same cluster as your VM, a local backup off the cluster but in the same region, and a DR backup on another provider in another country. Each one survives a wider failure than the last.

Failure Snapshot Local backup DR backup
Bad migration
Deleted file
Dead disk
Dead node
Dead cluster (SDS)
Dead datacenter / region

Same data. Two very different failure domains.

A snapshot sits on premium replicated cluster storage next to your VM. A backup is compressed, encrypted, and shipped to object storage off the cluster: the local S3 for a cheap extra copy, or an independent provider abroad for real DR.

Backup Snapshot
Failure domain Off the cluster
local S3 or independent provider
Same SDS cluster
Price per TB / month €35 local S3
€70 off-site DR
per copy, summed across backends
€35
premium replicated storage
Billed on Compressed archive size Full provisioned disk size
Backup type Full self-contained copy every run, no incremental chain Full point-in-time image
Copies One per selected backend; DR backends are in different countries One, on the cluster
Verification Optional boot-test, €15/TB restored per run n/a
Where it lives Local S3, or another provider in another country, under your key Same SDS cluster, three live replicas
Survives Cluster loss; a DR copy also survives the region and the provider Disk, node and hypervisor failure
Restore speed Slower: data is pulled back and decrypted Fast: rollback inside the cluster
Best for Disaster recovery Operational recovery

Not sure which one you need? Read Snapshots & Backups in the docs, or the blog post Most people think they have backups →

Example monthly cost

Backup cost tracks the compressed archive size. Backups and snapshots price on different bases, so they are shown apart.

A 20 GB compressed archive kept as a nightly backup on the local S3 runs about €0.70 a month. Put that same copy off-site in another country and it is about €1.40.

Compressed archive Backup, local S3
€0.035/GB per month
Backup, off-site DR
€0.07/GB per month
5 GB€0.18€0.35
10 GB€0.35€0.70
20 GB€0.70€1.40
50 GB€1.75€3.50
100 GB€3.50€7.00
250 GB€8.75€17.50
500 GB€17.50€35.00
1 TB€35.00€70.00

Snapshots price on the full provisioned disk, whatever the disk holds:

Provisioned disk Snapshot
€0.035/GB provisioned, per month
20 GB€0.70
50 GB€1.75
100 GB€3.50
250 GB€8.75
500 GB€17.50
1 TB€35.00

Multibackup: each copy is billed separately at its backend's rate. A 20 GB archive on all three backends (local S3 + two off-site DR) runs €0.70 + €1.40 + €1.40 = €3.50 a month, and lives in three countries at once.

What a single backup costs

Billing is hourly, so you only pay while a backup exists. A 20 GB nightly backup on the local S3, by how long you keep it.

Kept 1 day

20 GB, one night

€0.02

Kept 3 days

20 GB, default retention

€0.07

Kept 1 week

20 GB, weekly point

€0.16

Kept 1 month

20 GB, full month

€0.70

Every point is a full copy, not an incremental delta.

There is no chain to reconstruct and no base backup a later one depends on. Any retention point restores on its own, even if every other copy is gone. The trade is that each point you keep is billed at its own compressed size, so retention cost scales with how many points you hold, at the backend rate of each.

A single small backup costs a couple of cents on the local S3. A full GFS set holds up to 26 restore points (7 daily + 4 weekly + 12 monthly + 3 yearly), and every point is a full compressed copy billed on its own. For a 20 GB archive that is up to about €18 a month on the local S3 once the set is mature, roughly double off-site. Early on you only pay for the few points you hold; the cost grows as the monthly and yearly points fill in.

Every backup leaves the cluster encrypted

Every copy is compressed and encrypted with your key before it leaves the hypervisor. Keep one on the local S3, or fan out to three countries for disaster recovery and boot-test each one.