And yet, if you look at the AWS Re:invent presentation Frontier gave, they make a point about how Scalable it;s supposed to be, using AWS Elastic components.
Another fine line with very little substance it seems...
Do you know how AWS works? If so you probably know not every piece of AWS infrastructure will scale on the fly. Things such as databases (RDS) won't scale based on traffic. If those transaction servers are running on RDS, then this is probably what failed.
If its about server (EC2) instances, they probably hit some maximum limit they have setup. AWS doesn't let you scale indefinitely, as this would probably make you bankrupt before an attack was over. And this was basically a DDOS attack.
Too many transactions at once (the jumps) and after that fell over it became a DDOS on the login system. I honestly don't think they could have done much about this, there are always limits somewhere you can hit somehow even with the best planning.
Respect to the people fixing this on the weekend.