↓ Skip to main content

“Therefore, on taking over the duties of the primary server, the backup system applied the same logic to the flight plan with the same result.”

James Haydonjameshaydon.github.io ↗

One valid flight plan with two waypoints both named DVL, about 4000 nautical miles apart, took down NATS on 28 August. More than 2000 flights were cancelled, at an estimated cost of over £100 million. Primary and backup both crashed in under 20 seconds because the backup ran the same code on the same input. Redundancy that runs the same bug twice is just a slower single point of failure.