Availability
I mentioned earlier that one of the big advantages of "serverless" (and in particular, S3) is that it takes care of a lot of the "IT ops" work that traditionally engineers at every company had to homebrew.
One of those is availability: how often your service is up and running, serving user requests. It's often measured in "nines" - like "three nines" (99.9%) or "five nines" (99.999%).
See, users don't like when they log into your web app and stuff isn't loading. They don't like to hear that you're "down for maintenance".
AWS and S3 aren't perfect - but they are really good at availability. When AWS has outages, it's big news. Partly because so much of the internet runs on AWS, but also because they're rare.
You could build your own cluster of servers with better than or equal to one of the large cloud provider's availability. But it's very hard, and very expensive.
What does 'availability' refer to in the context backend engineering?