Yeah totally. We use GitHub + actions for backend, and self host Perforce + TeamCity for our main game codebase. We had 0 downtime in 12 months on our TC as a comparison. I know there’s a difference between running an internal service and developing a global scale platform that is abused, but as a user I don’t care about those concerns, I care about the platform being up!

"You can't possibly hope to have as many resources caring about uptime as $provider, you should outsource to them"

"Give $provider a break, they have such crazy scale that they can't possibly hope to have great uptime"

... yet it very rapidly gets lower uptime than a service running on a desktop in the corner of the office with some backups that get restored somewhere else.

Most sysadmins will tell you tales of laptops with a decade of uptime hosting simple services that nobody cared about (IRC, ticket software) with no downtime, not even an hour, and people only discovered that fact when they decided it was too slow and it's time to migrate.. These services have become less reliable than that, and servers themselves have only gotten more reliable in that time..

(yes, I'm aware of the security liability of decades old software running, even if it's not accessible by everyone)

There's a weird doublethink going on.