Every layer has failover
Servers: clusters and VM reboot/live-migration when a host fails. Network: redundant links, VRRP gateways, and dynamic reroute. Storage: multipath I/O, RAID, and replicated arrays. Power: dual-PSU A/B feeds, UPS, and generator transfer.
Unwatched redundancy rots
The most dangerous state for infrastructure is not failure — it is silently running on backup. One PSU died last month; the other is carrying the whole device alone. A storage multipath leg dropped; nobody noticed. A standby node deteriorates while idle. Key advantages: detect "running on backup" states, alert on redundancy loss before business loss, continuously verify backup health, track A/B power balance, and run failover drills with confidence.
FAQ
Failover is the automatic switch from a primary system to a backup when the primary fails. Failover vs high availability: high availability is the goal (services stay online); failover is one of the mechanisms to achieve it. Active-active vs active-passive: active-passive lets the backup sit idle; active-active runs both nodes simultaneously, sharing load and taking over from each other on failure.