Replacing an airline scheduling system sounds insanely complex. So many moving pieces, tens of millions of physical entities to track per day: crew, passengers, planes, spare parts, luggage, crew maximum allowed shift schedules, open gates, interacting with domestic+international terminals, seat assignments, and so on. Edge cases on edge cases on edge cases developed over years, probably in some god awful language running on a mainframe somewhere before “best practices” were a thing.
Modernizing such a system would take enormous political + real capital. The new system would undoubtedly have many growing pains as formerly resolved problems were not covered in the new system. It’s no surprise nobody would want to touch such a system.
One of the more tangible outcomes from my work on mobile sync is the United scheduling software. Seems like Southwest was too little too late on IT upgrades. https://www.couchbase.com/customers/united-airlines/
How is Couchbase in general and the mobile sync in particular relevant to United not having this disaster right now?
I've worked with airline scheduling software at almost the scale of United (about half the size, for the largest customer). That was using an Oracle database, which was a bit of a pain and a big expense, but worked fine as long as there was competent admins and competent devs. Would Couchbase be disaster-proof even when run by clowns?
Not saying there aren't better and worse choices for databases in any given situation. Just saying that there are lots of them that work perfectly well in competent hands.
Architecture advantages over the phone based system described upthread are obvious. The reason United chose Couchbase is because it’s designed for offline updates. So if a plane, phone, or airport is disconnected they can still do the data entry work and reconcile asynchronously.
But the Southwest system is not "phone-based" by design, I presume? They are going via phone because the decision support system for ops/crew control are down/unusable?
If they've resorted to doing all their scheduling manually with pen and paper, any data pushed to the edge (air crews, ground staff etc) will need to be transmitted in a similarly manual fashion.
The gate/ground/air crews are not making autonomous decisions about who is flying what, when and where. That is decided by some ops/crew control structure in the airline.
Modernizing such a system would take enormous political + real capital. The new system would undoubtedly have many growing pains as formerly resolved problems were not covered in the new system. It’s no surprise nobody would want to touch such a system.