0. Think of an array as a run-flat tire - it will let you hobble along after a blowout, but you want to get the thing fixed as soon as possible: having some drives alive when one dies lets you bring your work to some conclusion instead of everything hitting a brick wall while you go get a new drive and restore from a backup (count the hours and hours... zzzzzzzzzz).
1. They are not a substitute for backups. If you have an array of drives in whatever configuration, you still need to make backups. Even if you never have a mere drive failure (which you will), relying on an array as an automatic backup means you're expecting never to have any other catastrophe: lightning strike, you scooting across the carpet and zapping the machine with your static-endowed finger, your kid fiddling with it, house fire, etc. that destroys the physical location the drives occupy. Even if none of those disasters should ever befall you, as a human you err sometimes: a human operator malfunction gets blown onto all the drives in the array. Oh, no backup? Whoops...
2. If you have an array, it's not a bad idea to avoid having any two of them from a single lot: they'd be manufactured at the same time, would be powered up for the same length of time, be accessed at the same time and... yep... quite very possibly fail at the same time - and then you're no better off than if you had just a single drive. Even if they don't fail simultaneously, you will find yourself replacing another shortly after its brother.
3. Avoid like the plague any drives whose manufacturer starts with 'S' and whose model is named after a fast cat or a mean fish.
4. If you *really* want to devote yourself to an array, have at least two controllers. Single controller RAID means you expect the controller will never fail but... Surprise! Also, when one controller does fail and blow crap data across its drives, your other controller lives on and does not blow crap data - leaving you some number of drives that have good data.
In any case, make backups, and then remove them from the building in which your workstation lives. Regardless of whether you use physical media or backup to the cloud, a backup stored right next to the computer is not really a backup at all.
Also consider that, in certain emergencies, cloud backups may be unreliable because your internet connection is down.
When you take a drive, make an image of the whole thing on another drive and then replace the first drive with the second, one of a couple things will happen:
1. The system boots fine and life goes on.- or -
2. Windows will boot to some point - maybe you'll get to the logo/loading screen, maybe not - and you'll then be stuck in an endless reboot cycle that only ever gets to that point.If you get scenario #2, and you're sure your image is physically okay, your Windows install may not have the necessary drivers enabled to talk to the new drive correctly. You sometimes encounter this if you've got a moderately older system and older drive, and suddenly have to swap in a brand new drive. While not impossible to rectify, it's definitely no fun, particularly because Windows doesn't tell you directly what the problem is and you have to figure it out without much info to go on. ...this happened to me recently with a client I was assisting; it was double-plus un-fun. I won't bore with the details of what to do to try and get around the problem, but I'm happy to provide them to anyone who winds up in such an unfortunate circumstance.