I can't wrap my head around how extraordinarily complex airline reservations/logistics systems are. I'm kind of surprised you don't see this more often.
UPDATE, 1:43 P.M.: The Federal Aviation Administration has confirmed the ground stop. “GROUNDSTOP IMPLEMENTED BY ATCSCC ON AMERICAN AIRLINES NATIONWIDE AND AMR AMERICAN EAGLE INTO DFW, LGA, AND ORD AT THE REQUEST OF AMERICAN OPS. STOP REQUESTED AS RESULT OF AAL SABRE SYSTEM FAILURE.”
Looking at Sabre's own timeline, I cringe at what I suspect the codebase must look like. From what I read between the lines here, it looks like neither a carefully maintained legacy system, nor a proper rewrite at any point along the way, but more like 50 years of things bolted onto each other: http://www.sabre.com/home/about/sabre_history
I have the utmost respect for systems that have been running for decades without major rewrites, but having said that, I imagine there is a rock-solid core that has, literally, withstood the test of time, surrounded by semi-compatible systems that accreted around it since the mid-80's, with most current systems only touching this peripheral mass rather than the core services.
I've had contact with many systems that have their cores running on IBM mainframes with peripheral systems running on .NET. I fear those.
United's failed in November, although for a much shorter duration (hour or so). As luck would have it, I was flying United that day, and I'm flying American today =/
It's not that hard of a problem to solve really. Weve come a long way as an industry since SABRE came out 53 years ago. AA's biggest problem is that their IT dept is 80% low grade contractors and they've been unsuccessful in delivering their own solution for what SABRE does. They've been pushing back the sunset of SABRE for over a decade. Many days I'm surprised anything written at AA runs at all.
