Someone in one of these threads said there's a hidden DNS within VPCs that can fail and isn't scaled, so if that's true, they might just have to architect around that unless they can get AWS to change it. It's on RH for not knowing that but it's also kind of on AWS too.

But as far as what you can do, you can really only split your cash across brokerages if you want to engineer the same redundancy yourself. Otherwise, RH would need to route everything to another exchange to keep satisfying orders, and even that is just another system that could fail. Keeping all of your money in one brokerage doesn't seem ideal if you want to completely avoid downtime. Doing the same redundancy yourself with those industries isn't really practical.

High trust systems require just that, high trust. And once broken it's hard to re-establish.

Crypto exchanges certainly have their fair issues of downtime, but don't forget that crypto exchanges for a long time operated purely for early adopters as crypto wasn't something that everyone traded. There was also less availability of competition, because again the industry was newer and there were fewer choices.

And certainly Coinbase helped to popularize crypto trading and they had their fair issues, but I don't believe they had an outage of this exact magnitude, and again they were in an early adopter area where mistakes are seen as part of the process. If not expressly, then at least subconsciously.