On June 1st, 2009, Air France Flight 447 took off just after 7pm localtime from Rio de Janeiro, bound for Charles de Gaulle airport in Paris. Four hours into the trip, the plane crashed into the Atlantic Ocean, somewhere off the eastern tip of South America. To this day, no one knows why.
Shortly before it disappeared from radio and radar, the plane's computer sent automated messages that something was wrong. The contents were an odd combination of unrelated systems, and strangely, that the airspeed indicators were in disagreement.
The end result is that we don't know what caused the plane to crash and kill 200+ people. The problem could have been related to a local weather condition, or pilot error, or a systemic problem with that model of airplane. We don't know, though, and that's the scariest part.
Flight 447 was an Airbus A330 (specifically an A330-200), and there are hundreds of those planes in service today, carrying thousands of passengers around the clock. The lives of the people on those planes may depend on us finding the "black box", the flight data recorder. Essentially, it looks a bit like an orange shoebox, and it's somewhere on the floor of the Atlantic Ocean.
The FDR has a battery-backed beacon that lasts for 30 days. While that does mean there's no chance that it's still pinging, we do have the original recordings from immediately following the crash, and responders were first searching. As the IEEE Spectrum is reporting, investigators have reanalyzed the recordings, and believe that they have the site narrowed down from 1,200 square miles to 40.
40 square miles would still be a lot, even if it were on land. That particular area of the ocean floor is mountainous and in some places, several miles deep. Recovering an FDR from that depth isn't without precedent, however. South African Airways Flight 295, which suffered a fire in the cargo hold which subsequently caused the plane to crash into the Indian Ocean, had its FDR recovered from over 3 miles deep. It is entirely possible that Flight 447's recorder will be found, now that we know where to look, and lives may be saved because of the contents of that recording.
By now, you may be wondering what this has to do with servers, networking, or IT in general. Directly, not much. The underlying technology behind the flight data records is actually really cool (pdf warning). But no, I'm not talking about the technology behind the planes themselves. I'm talking about the methods, the drive, and the sheer determination to discover, at all costs, the root cause of the issues that occur in the aviation profession.
I have frequently thought that there are few disciplines with higher uptime requirements than IT. Accountants don't typically get called at three in the morning because the spreadsheet doesn't balance. You don't often hear of a 24 hour emergency insurance agent, and you almost never see postmen after sunset. But the people who run IT are called upon to constantly ensure that their infrastructure runs. The exact same thing could be said of aviation engineers, with the exception that when sendmail crashes, people don't die.
There's a certain amount of increased incentive to ensure that aviation accidents don't happen, and when they do, the problem is identified, analyzed, and fixed. There is no "reinstall and try again" on an airplane. If your flight control system fails and you don't have a backup, bad things happen. So you have a backup.
As I mentioned, when problems happen in IT, they're often significantly less dire. However, we do spend a large amount of our resources on ensuring data integrity and service continuity. Perhaps it wouldn't be unwise of us to look at airline engineers and safety policies as being a model to work from. As far as I can determine, they did invent the concept of a checklist for complex tasks, and their success to failure ratio far exceeds mine, at least in terms of accidents per hour. I'm not suggesting that we go to the lengths that the National Transportation Safety Board (NTSB) goes to, but we can always use additional reminders to be diligent in our tasks to protect data and prevent loss.
If you're interested in more comparisons of IT to aircraft engineering, check out my contribution to SysAdvent: Day 13: Redundancy.
- Log in to post comments