4 ms·
> Distributed systems are usually more robust against disruption that centralized ones Distributed systems can also be more complex to re-converge after a majo
by ethbr0 5y ago
> Distributed systems are usually more robust against disruption that centralized ones
Distributed systems can also be more complex to re-converge after a major disruption.
I'd imagine it's non-trivial to re-plan a network, given equipment and crew limitations.
And as context, we're talking about 24 hours later, now?
- bobthepanda 5y agoYeah. Part of Southwest‘s high utilization rate is their ability to turn around planes and crew quickly. In the event of widespread disruption, planes will not make it to their next airport to become a different flight, nor will their crew. Crew can also only be awake and working for a limit of time, so if the disruption causes a crew to go over that limit they are not going out until the next day, etc. With cloud computing, it‘s a bit different because which host you‘re using in a region doesn‘t really matter, but for airline a specific plane with specific people need to be in a specific time and place.
- smsm42 5y agoBut not using the same plane would not make it any better, neither would using a hub - that's my point. This is a common problem, but the properties outlined in the parent post which are supposedly specific to SW do not make the problem worse.