Redundancy is duplicating critical parts of a system so that no single failure stops it. In surveillance that spans the whole chain: RAID so a disk failure does not lose footage, a spare (N+1) recording server so a server failure does not stop recording, dual power supplies and UPS so a power fault does not black out the system, redundant network paths so a switch or cable failure does not isolate cameras, and edge recording so a network outage does not create a gap. Each layer removes a single point of failure.
The reason it matters in surveillance specifically is that failures and incidents correlate. The storm that cuts power, the break-in that damages a cable, the overloaded night when everything is under stress — these are exactly the moments the footage is most valuable, and the moments a non-redundant system is most likely to fail. Redundancy buys the assurance that the record survives the event it was meant to capture.
The pitfalls are partial redundancy and untested assumptions. A system with RAID but a single power supply, or dual servers but one network path, is only as resilient as its weakest unduplicated link — redundancy has to cover the whole chain to mean anything. And redundancy that is never tested often fails when called upon, so failover and backups must be exercised, not assumed. Finally, redundancy is not backup: RAID and failover protect against hardware failure but not against deletion, fire, or ransomware, which need a separate, ideally off-site, copy. Plan redundancy end to end, test it, and pair it with real backups.
