I dont have rally any info or anything specific to say about the not situation, but I can tell you that systems like these have a tendency to fall like dominoes.

You see, often, these sites are relatively near current capacity with room for fail-over overhead from somewhere else. But, if you have a really big failure or multiple failures, things can go over capacity trying to cover for the stuff that went offline.

With networks and most things in the IT world, going over capacity often snowballs into a negative feedback loop, where the maxed out things are too slow, so things time-out, try again, people get impatient and press “refresh” 73 times, etc. This can multiply the already maxed out traffic, keeping things basically DDOSed for even days while it gets sorted out.

On top of that backups that people have set up are often poorly tested and dont actually work when its time to need them, like forgetting to replace the expired fire extinguishers.