Originally Posted by appleguru

I don't think anyone is saying it is trivial... But I also don't think it needs to be that hard either. I see a few possible solutions to this problem (I am by no means an expert here):

1) DB failover: redundant DBs that automatically replicate and self-elect a master. Requires 3 or more members so you can have a quorum during master election. During normal operations, all reads and writes happen to an appointed master instance. All instances send "are you still online?" Heartbeat messages to each other to verify status, and slaves replicate data from the master as soon as it is written. If the master ever goes offline, the remaining members vote and elect a new master, automatically resuming operations.

2) blockchain-like distributed ledger. Biggest problem I see with this I current implementations is significant time delay while transactions are verified (proof of work currently takes lots of time, by design). But on the upside, even "untusted" parties can help contribute. Doesn't make a lot of sense for an airline.

3) fully redundant DBs: similar to a raid disk array. Writes happen across multiple DBs in parallel. Writes are verified for congurancy across all mirrors when a read occurs. If a mirror drops, the set is degraded, but still remains functional until the mirror comes back online. Biggest downside to this approach is the additional time needed to verify reads, which may not scale fantastically. But then again, we're talking milliseconds... If you have to wait a few extra ms to confirm that seat/ticket/weight balance call/whatever, it's probally not the end of he world.