I'm no expert on this, but I do have experience with mass storage and backups and maintain large libraries and having redundancy and avoiding down time, so maybe I can at least get you started in some directions to do some more research in.
First you need to figure out what level of redundancy you need. You also need to know how things actually work. For example, you mention if a HD failed, it not affecting current workflow, but that's only even an option if the HD that failed was the drive the OS was working from and/or the source drive of the media being edited and saved to, outside of RAM. If one of those fail, you will lose current work. Saved work will be fine likely. But current work sitting in RAM will dump and be lost. These systems are not designed to have physical drive failure and dump memory to ... the failed drive for safe keeping? See what I'm saying? Kind of is one of those problems. Just like if the power went out mid-edit and you were not on a backup battery. Whatever was in RAM at the same is lost. Saved files are fine though. Something to think about.
NAS is network attached storage. I wouldn't stress this right now. It's not a backup really. It's just another way to work. A NAS is just a way to have storage over a network, so that you can avoid messing around in your working computer, or have a place to store data and/or even work from that is accessible by multiple machines/users if needed. You can make a NAS complicated by adding redundancy to it, but really, again I would not push into this. You can build for it and still fail and still fail to rebuild the array and still lose data and have that down time and no data at the end of the day. A multi-drive NAS with redundancy and no down time sounds dreamy, but it's far from perfect. It's not a replacement for an external physical copy off site, or external digital copy like cloud storage off site.
RAID is just an old, old way to take small hard drives that are not expensive and string them together and use them as a single medium. It's risky. Even with redundancy, like RAID 1 (mirror, 1:1 copies), you can still have total failure and lose data (as it's in the same machine, on the same controller, doing the same tasks at the same time). And going into more dynamic RAID arrays, like 5, isn't really worth it either now, as it's even more of a chore to rebuild if there is failure and you still can easily lose the whole array of data. I would just avoid the idea of RAID all together.
Cloud storage is modern. It's not terribly fast, well, for everyone. But it's a great way to avoid total loss and is off the physical site. That's a good property of a good backup. It's not the only way to go, it's just another tool to add to the mix so that you have a diverse web of redundancy so that if one fails, you have others keeping some data intact. This is where you keep important current data that you cannot afford to lose. This is not where you keep vast libraries of 8~10 year old wedding RAWs that have not been culled.
If you really want redundancy, I would just have separate physical copies at different physical locations and/or with a cloud service too. It's not bullet proof, but there's no rebuild times, no special equipment, no real maintenance other than keeping a schedule to backup to all locations, rotate drives to keep healthy new drives in service, and learning a really important thing: culling data. You don't need to save RAWs from a wedding you did 8 years ago. It's done. Gone. Move on. Delete the data. You have small final JPGs that you can reference if needed, forever, somewhere on the web if you wish. You don't need to invest money into harboring ancient data that serves no purpose. This is for work, for weddings, not cataloging critical data. Try to keep the need in line with the reality of the usefulness of having the data for a long time, ie, more than 3~4 years.
Instead, I would focus on limiting potential down time. For example, if your main editing machine's HD/SSD with the OS on it failed, how do you respond? Do you replace it, reinstall all the software again all over on a new one and get back up and running? That takes a lot of time. Instead, if you had imaged the hard drive or SSD after you had it setup the way you wanted for editing, you could then easily just apply the image to a new drive in a few minutes or an hour or so, turn it on, and be right back up and running without having to configure or do anything. This is a smart way to backup your working machine. Hard drive cloning or imaging basically. It has nothing to do with the wedding images though, and everything to do with keeping your editing machine up and running with minimal down time in case of a HD/SSD fault.
Keep working medium off of the hard drive that your OS is on. That way if the medium drive fails, your working OS machine is fine; likewise, if your OS drive fails, your medium is intact on another drive.
Consider not using hard drives at all in the working machine and for when working on. Hard drives can be good as massive backup tools for physical copies. But hard drives are physical drives that move and have spinning parts, get hot, have heads inside, etc. Instead, consider using higher capacity SSD's that do not have moving parts. Less likely to have catastrophic mechanical failure. Plus they're much, much faster which supports your work flow.
Remember that if your electricity goes out, you can lose work. Everything goes down. Has to come back up and get back to work. Having uninterrupted power supplies can be a life saver and keep your uptime up, or when it's a total black out, at least let you save your current work and shut down and wait it out.
Lastly, having a 2nd machine is always a good idea. It doesn't have to be top shelf. Just something that will do some work in a pinch, while your main machine is down. If editing is how you make a living, the last thing you want to have to do is stop editing, leave, purchase a new system or get new parts, take the time to work on it, etc, losing lots of work time on a deadline. Weddings are not something to toy with after all. Having a 2nd machine, nothing fancy, a cheap beater, is a good way to keep working at least to some capacity while your down time on your main machine isn't costing you a client or income.
Hope that gets you started on where to read next.