DR backups
DR Backups
Keep several copies in different jurisdictions, with an optional check that each one boots.
What a DR backup is
A backup is a full copy of a disk kept away from the disk itself. DR stands for disaster recovery: the copy is still there when the server, the storage cluster or the whole provider is gone.
Ours are encrypted copies of server and volume disks, stored off the cluster. A DR copy goes to another provider in another country, and one backup run can write copies to several places at once.
Why ours
Encrypted with your key
Every backup is encrypted with your account's key before it leaves the host. You can copy the key in the Cloud Panel and keep it yourself.
Several countries in one run
One backup can write a copy to several backends in different countries at once. Each copy has its own checksum and its own restore, so losing one provider leaves the others untouched.
A check that it boots
Turn on verification and after each backup we restore it into a sandbox server, boot it and wait for the guest agent and SSH to answer. You learn that a backup does not start on an ordinary day, instead of during an incident.
Checked on the way in and out
A short or broken upload fails the backup instead of saving a bad copy. On restore the disk is checked against the hash taken at backup time before a server gets it.
Restore and retention
Restore a server in place, where its disk is replaced with the copy and the server starts again, or start a new server from the backup. A volume backup restores to a new disk.
Automatic backups run daily in a time window you choose. Retention keeps the last 3 days, or 7 daily and 4 weekly backups, or 7 daily, 4 weekly, 12 monthly and 3 yearly. Backups you make by hand stay until you delete them.
Each backup shows how old it is. A verified one also shows how long its restore and boot took.
Snapshots
A snapshot is a copy of a server or volume disk at one moment, kept next to it in our storage cluster. Roll back in place or start a new server from it.
Take one before an upgrade, a migration or a risky deploy. Snapshots also run daily or weekly on a schedule that keeps the newest ones, and the ones you take by hand stay until you delete them.
A snapshot survives the failure of a host, because it lives on replicated storage. Only a backup survives the loss of the cluster.
Where copies go
ServersCamp S3
Romania. A local copy in the same region as your servers, the fastest to restore.
Wasabi
Germany. A DR copy at another provider in another country.
ImpossibleCloud
France. A DR copy at another provider in another country.
Questions
What is the difference between a snapshot and a backup?
A snapshot is a quick way back, kept in the same storage cluster as the disk. A backup is a separate encrypted copy off the cluster, and it is still there when the cluster is gone.
Does a backup stop my server?
No. The disk is read at one consistent point while the server keeps running.
Are backups on by default?
No. Turn them on for the whole account under Backups, or for one server in the create wizard or on its page.
Which servers can be boot-tested?
Linux servers from our templates, which boot with cloud-init. Servers installed from your own ISO and Windows are backed up and can be restored, but they are not boot-tested.