Microsoft once defined "uptime" as including time spent rebooting due to patches and necessary updates. So even though the systems weren't actually available, because they were being rebooted due to a deliberate event we were told that they were considered "up".
Utter and complete bullocks of course, but that was their definition a few years ago at least, when I last worked with Microsoft servers. I guess it was the only way they could reach their advertised 3-nines uptime.
Are they doing that for their cloud offerings as well? How IS Microsoft re-defining "uptime" these days?