Single Point of Failure: How Can One Small Part Stop a Whole System?

A single point of failure is not simply any component that might break. It is a component whose failure leaves an essential function with no alternative path. If every computer in an office connects through one network switch, the switch looks ordinary while it works. When its failure disconnects the whole office, its structural role becomes visible.

This differs from a bottleneck. A bottleneck limits throughput while the system is operating; a single point of failure concerns whether an essential function stops when something fails. Adding a second device does not necessarily remove the risk. If both devices share one power supply or one cable route, the apparent alternatives can fail together.

Identifying a single point of failure therefore requires mapping dependencies, not merely counting backups. Genuine redundancy must be sufficiently independent and its changeover must be tested. The concept replaces the vague claim that a system “looks reliable” with a sharper question: which one failure would leave the essential process with nowhere else to go?

https://csrc.nist.gov/projects/risk-management/about-rmf