Call me paranoid about NVMe, but I would find a way to work in one big ass Sata HDD to, if nothing else, just store a nightly image of the other 4 drives on... Just in case.
As you should. While SSDs have no mechanical parts, they are more prone to failure than a HDD, and in a different, nastier way, too.
The flash cells (the things actually containing the bits and bytes) themselves are organized in blocks (usually called pages) around 4K to 16K. When there is nothing there yet it is not a problem, the controller can directly write to that part. But when there was previous data that needs to be updated (think about adding a file to a directory), then there's a small issue: Flash cells can only be written from neutral state (no electrical charge, reads as a 1 bit) to a charged state (0 bit) once. If you want to write a 1 bit (no charge) where currently is a 0 bit (charge), you need to reset the flash cell to the "no charge" state.
But erasing a cell isn't that easy, because of the sheer amount of flash cells they can be erased page by page. So if you update only 1 bit in a 512 byte sector, you suddenly have to reset the complete page (4K to 16K), and then rewrite it's data with the changed information.
But this nifty trick can only be done so many times, and this is the reason modern SSDs have something called "wear leveling", were they distribute that freshly written data to page that hasn't been erased as much as the other pages. How often this can be done depends on the cell technology: single level cells (stores 1 bit) up to 100.000 times, quad level cells (stores 4 bits per cell) between 500 and 1000 times.
Also, because SLC can be written faster, as an additional trick up their sleeve modern SSDs now store written bulk data to a part of the SSD as SLC information, and then once writes settle down rewrite it from that SLC cache to TLC/QLC. This is why some SSDs can write real fast up to a certain point, and then suddenly stall: It's not about heat. In fact, keeping the flash cells nice and toasty prolongs their life when writing. Their SLC quota is simply used up, and now they have to write directly to TLC/QLC.
Rotating hard disk drives are prone to mechanical issues, but if you have a competent data rescue lab parts of the data can usually be recovered by reading the platter. When an SSD fails it usually loses it's map where the logical sector (from the perspective of the OS) was stored in which flash page on the SSD, and due to wear leveling this looks completely random. You may recover individual pages, but for everything bigger than a flash page you just have a million pieces jigsaw puzzle, except with flat edges. if you have a self-encrypting drive (SED), then you are just plain fucked.
tl;dr: Back up your stuff. Even the things you deem unimportant now. Rule of thumb: If it took you longer than 10 minutes to do, and 5 minutes to think about, then it may be worth backing up. If you can, do this to at least two different drives. Test your restore.
Haus
, I'm 100% self taught and only really care about PC's for playing games. So take my $.02 with a giant grain o' salt.
Main one
Worth checking
