This past week, Carol & I were in Ireland for the OpenStreetMap State of The Map conference (small, but interesting). However, not everyone who was supposed to be there showed up. Apparently, Thales (large French aerospace company that makes Magellan GPS devices) had put in a €100M ATC system in Dublin over the past few years and it was having severe problems.
The problem was tracked down by the manufacturer and [ttributed to a bad NIC (network card) in one of the computers involved in the Air Traffic Control network.
Disingenuously, the manufacturers asserted to the IAA (the Irish equivalent of the FAA) that the intermittently malfunctioning network card was "consequently overcame the built-in system redundancy."
Sounds nice, but the reality was that this problem persisted for nearly 5 days. Although the system may have "worked around" the problem using redundant equipment, the bad equipment was never removed from the mix, causing airplane information to disappear from controllers screens for 10 minutes at a time at random intervals. If the system redundancy had been appropriately handled, the intermittent failure would have resulted in both the flakey equipment being removed from operation and an alert being sent to the maintenance folks telling them what to remove.