Scheduled, with verified restores
Backups that are actually opened and restored on a schedule. A backup nobody has restored is a guess, and you find out which on the day you cannot afford to.
Products › Automation
System
Automation
The jobs that have to run whether or not anyone remembers them — and have to prove they ran. Backups, replication, monitoring and configuration history, set up once and then left to work.
The principle
Every manual routine is
a future incident report
A task that depends on somebody remembering it will eventually be forgotten, and it will be forgotten on the worst possible day. These are the routines we take off people.
Not stored beside the original
A copy kept on separate hardware, so losing one machine does not take the live data and its only backup at the same time.
Standby kept in step
Production replicated to a second machine on a short interval, so recovery means switching over rather than rebuilding.
Automatic takeover
Configured so a hardware failure brings services back on the standby without waiting for someone to notice and log in.
Health checks with escalation
Alerting on the things that precede an outage — capacity, link state, service health — routed to whoever is on duty.
Versioned config backup
Every device and server configuration captured and versioned, so any change can be traced, compared and reversed.
Why we are strict about this
Scheduled backups pass their own checks right up until the day you need one. The failure is almost never the schedule — it is that nobody ever opened an archive, or the copy sat on the same machine that died.
So we treat a failed restore as an incident in its own right. Finding it during a drill costs an afternoon. Finding it during an outage costs the business.
What you get
- Schedule
- Backup, replication and check intervals matched to how much work you can afford to redo
- Evidence
- A report that each job ran and what the result was — not silence you have to interpret
- Drill
- Periodic restore tests against the real archive, with the outcome recorded
- Runbook
- Written recovery steps, so the fix does not depend on one person being reachable
Works on your existing servers
When did you last
restore a backup?
If the answer is "we have never had to", that is worth checking before the day you do.