Thank you for your time and thoughts. Good stuff. I took you up on your offer to talk about drive types, etc. and messaged you.
For the others here, I'm also considering using my 2010 Mini Server as NAS. Is it viable for that purpose? I may get an 8TB WD Easystore shuck to put in there. If I do that, should I also max the RAM at 16 and put an SSD in? Figure that would be cheaper than buying an NAS unit, and possibly perform better.
Network Attached Storage can cover a very wide range of products and use cases. In a literal sense you could use a Mac Mini with a single drive as a NAS but it wouldn't offer redundancy or speed that better NAS options would. That may be fine for your needs though. You'll have to think through what you need for performance and feature sets. Are you trying to have a local backup and don't care about speed of data access? Are you looking for storage to use with 4K video editing? Probably something in between but there's a wide range in there.
Performance of storage is affected by a lot of parameters: network/transport speed, processing speed of the device, caching capabilities of the device, and speed of the array (which in turn is dependent on the speed of the disks and overhead of the protocols).
Network/transport: you need to get your data across some sort of cable to get to and from your NAS. Typical wired networks for most people these days a 1GbE (one gigabit ethernet). Note that's "bit" (denoted by the lower case "b") not "byte" (which would use an upper case "B") so divide by 8 bits in a byte to get the transfer speed in bytes (which is what we typically think about). So 1GbE is theoretically 125 GB/s (actually less because there is overhead in the network protocols themselves). If your NAS is just connected via typical home wired connection you are already down to the speed of a fairly slow HDD. Better NASes support 10GbE, that could get you 1.25GB/s (again back out some for network overhead) but would require that the rest of your network have some support for 10GbE too (any computer wanting to access at that speed must also have 10GbE support, or you could have multiple machines access with 1GbE each maxing out at 125GB/s but collectively getting close to 1.25GB/s, you'll need network gear [router/switches] capable of handling 10GbE as well and cabling [cat6] that can handle the higher frequency). There are NAS that support port aggregation as well and could have two or more 1GbE (or multiple 10GbE) cables that can be joined into one virtual connection, this can give you a 2GbE aggregate connection etc... Your router/switch needs to support aggregation too for this to work and again if the computer side of the equation only supports 1GbE you'll be limited on that side. Some WiFi APs have theoretical speeds in excess of 1GbE but those are highly environment dependent (is there a wall between the AP and device? how far apart are they? what's the total load on the WiFi at any given time?) so I would stick with wired. Some NAS also support more exotic network speeds (40GbE, 10GbE over SFP+ with fiber, etc...) but most likely you'll either have 1GbE, 1GbE with port aggregation, or 10GbE available to you.
Transport doesn't have to be network either, many modern NAS units support some form of DAS (direct attached storage) in addition to ethernet. DAS can use faster protocols than ethernet, such as Thunderbolt 3 (40Gb/s = 5GB/s). DAS will generally require a very short cable run as the protocols they support don't hold up over long distances. I'll stick with TB3 since that would be most common for the Mac world. TB3 will operate at full speed for about 1 meter (possibly 2). The cabling is expensive relative to cheap cat6 or fiber, you'll spend $80 for a single TB3 data cable. You'll then need your NAS to sit very close to the machines it is connected to. If you have one primary system you need to be fast though (or several depending on the number of TB3 interfaces your NAS has and they all are within a meter of each other) you can have very quick transport to the NAS and then network access to any system further away.
Speed of the array: Most NAS units will use some form of RAID to group multiple physical drives into a single (or multiple) array(s). Let's say you have plenty of bandwidth to your NAS, if your array is slow you can't saturate the bandwidth anyway so it just goes to waste. For example a TB3 connection (capable of 5GB/s) to a single 5400 RPM HDD will only operate close to the speed of that drive (about 100MB/s). Array speed is typically a factor of the speed of the underlying drives, the number of drives in the array, and the protocol (RAID level) being used. Read and write speeds are generally very different in RAID due to mirroring or parity writes so this requires some understanding of the various common RAID levels...
RAID 0 (striping) - this writes a piece of your data to each drive in the array. It writes to all disks at the same time and reads from all disks at the same time so you basically get to multiply up the speed of an individual drive by the number of drives in the array to get your speed, it is very fast. It's also risky, if you lose any single drive in your array you will lose the ENTIRE array, since your files were written across all disks you'll have effectively lost a percentage of every file you have. Unless you are purely looking for performance and don't care about redundancy this is not a good choice.
RAID 1 (mirroring) - this writes your data to every disk in the array simultaneously, so every disk is a mirror of every other disk in the array. When reading it can read a piece of the file from every disk simultaneously too. For writing this means that the speed is exactly equal to a single disk, if you have two disks you are writing double the data (once to each) at double the aggregate speed so 2x speed/2x data = 1x disk speed. For reading it reads at the aggregate speed of the array, so two disks would be 2x speed / 1x data (you don't have to read it twice) = 2x disk speed. If you a drive in a RAID 1 array you can simply replace it and the array will rebuilt from the other(s) in the array. RAID 1 gives a good balance between robustness (you can lose a drive without losing data) and speed (quicker reads, no write advantage), it's a good option for small arrays (just two drives).
RAID 10 (mirrored striping) - RAID 10 can be thought of as RAID 0 across a series of RAID 1 nodes. Typically you have a series of two disk nodes in RAID 1 and the stripe across all the nodes. You get the robustness of RAID 1 in that you can lose any single disk in any node and still rebuilt but some speed advantages to writes from striping. Let's say you have six drives in RAID 10, you would almost certainly have three nodes of two disks each behaving as RAID 1 in each node and striping across the three nodes. Each stripe gets written twice (once on each half of each node) so your write speed is 6x speed (six drives) / 2x data = 3x disk speed. Read you can leverage reading a piece of the file from every disk so 6x disk speed. Robustness is slightly less than some parity methods (below) because if you lose both drives in a single node you will lose the entire array (two drives across two nodes would recover though as each node could be rebuilt independently).
RAID 5 (single parity) - Parity RAIDS write data across all disks in the array but also write some parity information across the array so if any disk is lost is can be rebuilt from the remaining parity data. A big advantage to RAID 5 is more efficient footprint of parity vs mirroring; this translates to more of your raw hard drive capacity being retained as array capacity (in RAID 1 and RAID 10 your raw capacity is cut in half, RAID 5 loses one drive worth so in a four drive array you lose 25% to parity data). Parity RAID can be quite robust and performant but has limitations (and parity versus RAID 10 can be controversial among storage people ;-) ). RAID 5 specifically (single parity) has reached its limit for large arrays, without getting too complex here with modern large drives it is very possible (even likely) that during a rebuild a second drive could have an unrecoverable read error and the entire array could be lost. That can be countered to some extent by using quality drives with low URE rates and smaller drive sizes but I consider RAID 5 deprecated and would only use RAID 6 (or higher) for parity. Write speeds can take a big hit with parity because you are (potentially) reading and writing data twice (four total operations) to calculate and write parity data. Simplifying a bit, in a four drive array you have 4x speed / 4x data (actually two read and two write) = 1x drive speed for writes. Reads happen across all spindles for 4x read speed in that scenario. There are exceptions to that if the system you use can precalculate parity (writing large sequential data and alignment with the strips and your controller supports it) but best to expect you'll have a very significant hit on writes; this and random IO is why there is controversy between parity RAIDs and RAID 10, despite this I typically favor parity RAID (although 6 rather than 5) for large sequential (most home uses will fall into this category) operations.
RAID 6 (double parity) - very similar to RAID 5 except it writes two sets of parity allowing the array to lose two drives and still recover. Doing so requires more operations on write (three read and three writes) so write speed degrades further unless you can precalculate parity (so for an eight drive array that would be 8x speed / 6x data = 1.333x disk speed for writes, reads are still across all spindles to 8x read). RAID 6 has a high chance of surviving rebuild even with modern large drives provided they are high quality (low URE rates, do not buy cheap desktop drives, something like Seagate IronWolf are good).
There are others too (RAID 7 triple parity, etc...) and many that are effectively deprecated and shouldn't be germane. I'd stick to these for any home user.
As you can see, if you want an array that will saturate a fast connection you need a lot of bays, fast drives, or both. Your Mac Mini with a single drive will always be limited to single drive performance at most (let's say you put a Samsung 2.5" SATA III - presuming your Mini is SATA III and not SATA II - SSD in there you'll be limited to 500MB/s even if you have a 10GbE network adapter for it and related hardware) and have zero redundancy. A two drive NAS (again just looking at the array speed) in RAID 0 would be 2x the disk speed in read and write so two of those SSDs would be about 1GB/s read and write. In RAID 1 it would be 500MB/s write and 1GB/s read (but have some redundancy and half the useful storage). A four bay RAID 5 with those SSDs would be 2GB/s read (presuming you have enough transport to use that) and 500GB/s write (possibly 2GB/s write if you can precalc parity). The same four bay with 210MB/s IronWolf spinning HDDs in RAID five would be 840GB/s read and 210GB/s write (again possibly 840GB/s write with precalc parity situations). Four drive RAID 5 you would retain 75% of your raw disk space as usable. Finally an eight bay RAID six with SSDs could read at 4GB/s (getting close to saturating TB3) and write at about 650MB/s (or again 4GB/s with precalc of parity data) - with IronWolf spinning HDDs that's 1.6GB/s read and 260MB/s write (or full 1.6GB/s write with precalc parity). Eight bay RAID 6 would retain about 75% of raw storage capacity.
So lots of moving pieces here between number of bays, RAID level, and drive type to try and meet your objectives of how much storage, speed, cost, and redundancy.
Cache: On top of all of this many NAS units also support a cache which can significantly reduce the performance penalty of parity RAIDs. Some units can provide a RAM disk cache which will saturate any connectivity for a small amount of writes (depends on how much RAM the device has). Others allow the use of SSDs (some even NVMe SSDs) for cache space and can get near to or achieve connectivity saturation.
Processing capability: assuming you have connectivity and array speed/cache that meet your needs we finally come down to processing power. The lowest end NAS units have very limited CPUs. They aren't often the bottleneck but CPU is used for parity calcs and for encryption if not hardware offloaded. If you encrypt at rest (everyone does right? OK maybe not, but we all should) and the encryption is done on the CPU rather than dedicated circuitry it can impose a significant load on the CPU and could become the limiting factor. My two cents, just get a NAS that has hardware encryption and make it a non-issue. Most modern NAS units will also do a lot more than just provide storage though too (run VMs, become media servers, even Plex servers, etc....) and all of that consumes processor and RAM so if you plan to utilize that sort of thing it is one more piece to the puzzle.
One last note, RAID is not backup. If you have your data both on your computer(s) and your NAS at least you have it in two places but remember that arrays - even though they have redundancy - can fail. Having your only backup physically in the same place as your original data doesn't protect you from fires or theft. Always have a backup and preferably off-site. Cloud storage is pretty dang cheap these days and many NAS units will automatically send your data to a could backup provider if you wish.
Sorry for the wall of text but this will get you started, I can answer more specific questions as you start to narrow in on your priorities and budget.
