Failover is the automatic switch from a failed component to a working one: a second recorder, a secondary network path, a standby server or a redundant power supply. The point is continuity of recording and access, and it is only real if the standby is tested, monitored and current with configuration.
Redundancy is easy to buy and easy to assume. Failover is the part that has to work without anyone watching, which is why it needs testing.
What can be made redundant
Recorders can write to a standby unit, switches can be paired, internet paths can be duplicated and power can sit on an uninterruptible supply. Each adds a component that itself needs maintaining. Decide which failure a site cannot tolerate before adding hardware.
Testing the switch
An untested standby is a hopeful spare. Pull the primary link or power down the primary recorder on a schedule, and confirm that recording, access and alarms continue. Check that the standby holds the current configuration and firmware, since failing over to an outdated copy can undo months of changes.
Related:Glossary: NVRGlossary: VMS