Hacker News Reader

sure, why not? I mean, assuming there isn't some nuclear attack that destroys all networking infrastructure. I'm not suggesting it's easy, but if you distribute it enough and make it able to work when other portions of the system disappear i see no reason you couldn't claim 100% uptime

View on HN

And then you you have a stupid software bug and the redundant physical systems are not helping.

Azure Storage is a available in a ton of locations and all of them stopped working because Microsoft forgot to roll out their patch to renew the SSL certificate. So they knew about the issue but rolling out the patch was not done.

Another example was when they introduced HTTP2 for CDN endpoints and forgot to test the cipher list with popular browsers and broke compatibility with Firefox ESR.

Reply on news.ycombinator.com