Anyone who thinks a "server" in MMO user terms is actually a single, physical server that looks and acts like your home PC, is seriously fooling themselves. =)
In most MMO's from around the XI era, each zone is one or more physical blade machines controlled by one or more world servers, which are themselves dependent on additional servers for chat, authentication, trade, etc.
It's not uncommon for "a server" in user terms to be 50 or more physical boxes, many of which have to be brought up and launched in a certain order. Not in mass parallel all at once. An hour or more to "reboot the server," is therefore completely reasonable. Especially if any one of those restarts doesn't go precisely as planned and the whole thing has to be started over.
It's vanishingly unlikely that XI is still running on the same physical boxes as when it was launched, so the problem isn't hardware, but more likely, overall system architecture. It's the sheer size of the operation and the length of the checklist that has to go into it.
These days, more modern MMO's with higher engineering budgets architect themselves with the understanding (in hindsight) that certain types of patches are more common in industry operation than others, and that scope and breadth of the world will always grow with time. Thus, they're engineered so that different types of patches and different types of outages will only effect a minimal amount of servers with the least amount of downtime. It's even possible for some simple numbers or script-related fixes, like tweaks to item or combat stats, to be done without a reboot at all.
XI (and XIV, as it seems they didn't learn much) comes from an older era, and so the whole thing likely needs to be bounced in series every time there's an issue.
