back to article Handling tech baggage: How American Airlines, US Airways merged IT

At the end of 2013, US Airways and American Airlines merged. But actually combining the IT systems of the two companies is expected to take from five to seven years, said Susanna Brown, managing director of operations technologies at American Airlines. The two airlines move slowly because of baggage: They have different …

  1. Peter Prof Fox

    Economies f scale...

    So all those people who according to the PR guff were afraid for their jobs, how many ACTUALLY, REALLY, FACTUALLY lost their jobs. There may have been a tidal wave of programmers employed to do some merging but the bottom line is fewer staff for longer hours on less pay. Hurrah for capitalism.

  2. Nate Amsden Silver badge

    no way

    They are applying kernel updates to 200 systems and rebooting them in 5 mins without significant impact to availability(otherwise they would talk about much more than 200 systems)

    1. Steve Davies 3 Silver badge

      Re: no way

      Have an upvote for being sensible.

      Having worked in the airline/airport industry I know that availability of systems is the top priority.

      At one big Airport served by both AA and US on the west coast, getting a baggage system outage (to apply patches) would need at least 2 weeks advance notice. Even a failover (on an active/passive setup) requires a lot of advance planning with the airport and terminal managers. The last thing they want is a Baggage system that does not work and with bags in places where humans would find it difficult to get to.

      being 'conservative' is the name of the game.

      Yes, using tools like Puppet can speed up things but TBH, that would be the least of my worries if I was running the operation.

      Then there are two other big cans of worms sitting waiting to burst open and that is

      1) Baggage systems are custom made for the terminal. That means the software is often very customised.

      2) These airlines will more than likely have baggage systems from different vendors. Vendors are often very reluctant to make changes to the software that has been accepted into use.

      Just saying that using Puppet saves 'n' hours is being very simplistic. The whole thing is far more complex that that and these people know it.

  3. John Smith 19 Gold badge

    " much an organizational change-management effort as anything else,"

    A lesson British Governments (and the British Civil Service) still don't seem to understand.

  4. Disk0

    "we can set this to go off at night and we get a report in the morning"

    That statement alone is so terrifying I don't know where to start.

    Setting things to go off, usually implies either alarms or bombs; the "Fire and Forget" mantra has apparently extended here from HR to IT policy.

    I would not like to be on the receiving end of the morning report looking like this:

    "Migration log: start time 0:00 Error: process initialisation failed. End log." Or words to that effect.

    I know managers like the idea of automating staff tasks, but being there is still the only way to make sure everything goes as planned. I guess overtime for real work is a big nono with Big Air.

  5. smudge

    No, no, no - it goes THIS way round!

    It's a long time - about 13 or 14 years - since I flew on American Airlines.

    At the time, their in-flight corporate videos - safety info, guide to Dallas airport, etc - started with a view of the Earth from space. With the Earth spinning in the wrong direction! Made you wonder how good their navigation would be.

    I haven't found a copy of it, but I did find a short discussion of it here, to show that I'm not making this up.

  6. yoganmahew


    Their mainframe data was migrated seamlessly over the space of three months with the US flight designator retired at the end of it. Not a puppet nor a makey-uppy methodology in sight :|

    1. yoganmahew

      Re: Mianframe...

      Of course, if I could spell mainframe... :\

  7. JaitcH

    Out of the AA Group Sabre Reservations ...

    had the sharpest programmers.

    But American split Sabre Res off and sold it to get some cash.

  8. JK63

    Without knowing anything about the architecture nor the application it's tough to make the determination that it can't be done.

    It is likely that in this environment they've built wide (safety in numbers) everywhere possible, with load balancing. This is a very robust solution, and you can take servers down (virtual or physical) without taking them out of rotation. The load balancer's job is to detect the service outage and remove it from the group.

    From a database perspective, Active / Active clusters can be built which would allow for a DB server to go down without any visibility to either the application or customers. There's also Oracle's RAC (and probably others) as an alternative beyond 2 servers.

    There are ways to skin the cat that would involve minimal disruption and go unnoticed outside of the people managing the systems.

    I suspect they have taken just a little liberty though. I'd bet hey have an engineer or two keeping an eye on things as a CYA while they are patching.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like

Biting the hand that feeds IT © 1998–2022