I assume the too-shy-to-reply downvoter thinks downtime is not a direct correlation to a problem with the webserver. While that's true, I was hoping to argue that the stack chosen is indicative of deeper problems such as stack complexity. A too-complex stack is more prone to break, which is part of the appeal of NginX, its simplicity. You could also prove Microsoft stacks are more or less likely to fail. For example .01% of Microsoft servers were offline last month, .009% of Apache servers were offline (> 1 hour) last month, etc. PS: If it wasn't obvious, I made up hypothetical percentages.