ELi5: When a server is taken down for routine “scheduled maintenance”, what is typically being done to the servers while they cannot be accessed?
In: Technology
Sometimes it’s literally OS updates, like a Windows update. The website is small enough to only have one server (not hundreds like big companies) so when that server is off or rebooting the website won’t work.
Sometimes it’s just easier to change stuff, like migrate data, when no users are using the website right now. As a rough simplification and very ELI5, imagine Reddit changing where on their servers they store posts and comments – while they’re moving all posts to a new location (database), tons of new posts and comments will be added to the old location so they’d never finish the move. Thus, services are shut down, then stuff is getting moved / reorganized / whatever, then services are made available again.
OS updates, reboots and other major changes to the system that can’t be done while it’s running can be the reason.
Ever wanted to move, delete or change the name of a word document while it was opened? You can’t do that, Windows won’t allow it. It’s in use and Word needs it in its current state, right at this moment. You’ll have to close Word, do your thing and then reopen the changed document.
This also works on much higher levels. I recently copied a database from an old server to its future successor. After it was fully copied, I tried running it on the new server – didn’t work. The old server was actively running the database, making changes WHILE I was copying the data, so I ended up with an invalid dataset (think copying a sentence word by word, and when you’re halfway done the sentence gets replaced by another. Now you can’t get the original second half of it any longer, and it just doesn’t make any sense). I had to actually turn the database on the old server off for a minute, in order to copy a valid and uncorrupted version of it.
Depends, but the term encompasses everything from updating software, installing security patches, upgrading hardware, migrating data, etc. Truly routine maintenance on mission-critical servers would include thorough diagnostic checks to uncover and thwart potential problems before they become actual problems.
In short, could literally be as simple as “turning it off and turning it on again” to “running full backups + swapping out hardware + updating or swapping out software + reinstalling data + thorough testing.”
On a very basic level, a server is a computer with an operating system like you use at home. They get updates that require it be restarted so this could be it.
They may also be upgrading software or services on the server that require nobody to be accessing it at the time
Bigger businesses usually have multiple servers using losd balancing and other technologies so that if one goes offline, the other is still functional so the user doesn’t see any downtime
Updates, patching, adding new features or potentially removing some. It really depends on the server and its purpose but typically whatever is happening tends to be performed during lower traffic times given loss of availability.
It’s not that they can’t be accessed, it’s that public access is turned off. The engineers working on the server definitely have access. It could be anything from hardware upgrades/maintenance, to physical relocation, to software upgrades. Most of the time software upgrades can be done transparently, but sometimes you need to do bigger upgrades that require restarts.
Latest Answers