This is mostly aimed at large cloud providers such as Amazon, Google, and Microsoft. They can use units like this and take them up for scheduled maintenance eve

This is mostly aimed at large cloud providers such as Amazon, Google, and Microsoft. They can use units like this and take them up for scheduled maintenance every x years. What happens to some of the individual servers between maintenance is of less importance. The infrastructure can handle the failures and they can just turn off the affected servers if they cannot fulfill any role.