We recently deployed a system (linux based on a windows server with vmware) on a huge infrastructure provider. Every week the box was going down, and we asked why - it's because they reboot ALL microsoft servers once per week, across the entire infrastructure. He claimed they were so unreliable it's good practice. WTF?

Uptime on my first linux production server box was over 4 years.

Yes, in all large shops a weekly/monthly reboot cycle is a requirement. Duly note that this requires several hours of overtime from sysadmins, adding additional cost.