Two Phrases That Don’t Belong Together: Software Bugs and Airplanes
The mere thought of air travel during the holidays is annoying enough to send most people running to their nearest bus or train station. The crowds, the lines, the delays, the zip-lock bags and 3 oz bottles of shampoo… but wait, there’s more!
Late last week, a five-hour computer glitch caused flight delays across the U.S. that were still rippling through the transportation system for most of the day. The problem was made worse by the fact that the National Airspace Data Interchange Network failed at both its locations — Atlanta and Salt Lake City. (Ed. note: I’ll try hard to avoid using the word “crash” in this post.)
Bloomberg.com had this to say:
The Federal Aviation Administration blamed a four-hour software failure for causing airline delays and cancellations across the U.S. The shutdown lasted from 5 a.m. to 9 a.m. ET after “a software configuration” malfunction today in Salt Lake City.
And The New York Times chimed in with this little bit of sunny news:
Flight plans typically consist of hundreds of alpha-numeric characters, giving the flight number, type of equipment, takeoff location and various intermediate points, with altitudes. On Thursday airlines were faxing flight plans to controllers, who were typing that data on keyboards, not quite hunt-and-peck but not nearly as fast as a computer would transfer the information.
And now the really good bad scary news: This isn’t even the first time that this particular system in this particular location has had major problems. In August 2008, big delays were reported after the same system ground to a halt in Atlanta. Those issues were blamed on — wait for it — a failed software upgrade.
This all makes me think back to Bill Gates’ demo of Windows 98, which crashed (oops, sorry… I tried!) so hard that it probably registered on the Richter scale. Sure, it made for good TV and a minor ‘red-faced’ moment for Microsoft’s genius leader. But in the end, it was nothing more than a PR faux pas.
The degree of mission criticality of a system should be proportionate with the testing and failover and stability for that system. And as more of our transportation and energy infrastructures depend upon software, we as software professionals need to take this more responsibility more seriously.
