How exactly do all these cost comparisons factor in the ability to quickly deploy hardware in 4 different regions in the world without having people at location to swap broken hardware or reboot an instance that is not available for whatever reason?