The DL380 servers are designed for 24/7/365 operation when properly installed. I think we have some DL380 G7's that were originally powered up when installed several years ago, and haven't been powered down since. Only reasons for powering those down would have been non-hot-plug capacity expansions (e.g. swapping memory DIMMs for larger-capacity ones or adding a second CPU if originally bought with just one). It's entirely possible that the day they will be decommissioned will be the first time some of them will be powered down since their initial installation.
When the system is powered up, there is a brief but significant spike of electrical and mechanical load. If a server has been running for a long time (years) and some component is just on the edge of failure from just the regular use, rebooting the server can often cause such a component to fail completely.
Therefore, I'd say: if you want maximum lifetime from the server, keep it running. But when you need to make hardware changes or something else that requires shutting down a server that has been running for years, first power down and reboot the server once before making any other changes. If there is a hardware fault, the reboot should make it apparent, and there will be no reason to suspect that the new changes might have caused the fault.
Of course, there might be other considerations: you might, for example, have a system that will automatically power down unused servers at times of low workload, to minimize power consumption.
On the other hand, just rebooting the operating system might be necessary more often, to allow the update of some core OS components, although modern OSs can update most of the OS without rebooting. If your servers are exposed to the internet, you should keep up to date with patches; if that requires rebooting to make all the new patches take effect, then you should have a reboot schedule that essentially matches your patching schedule.
MK
