I’m a big fan of Backblaze’s failure statistics. https://www.backblaze.com/cloud-storage/resources/hard-drive-test-data

Annualized failure rates go from 0.3%/year to 3+%/year, even just looking at the drives they have million+ hours for, and I’d rather be at the lower end of that 10x range.

As you are looking for bulk data storage, the drive’s speed isn’t of too much concern. A 5400RPM drive is plenty.

If you are looking to put this drive into an array with other drives, make sure you get a CMR drive as SMR drives can drop out of arrays due to controllers finding them unresponsive. If a drive does not list it is CMR, it’s best to assume it isn’t. Seagate has a handy CMR chart, for example.

Additionally, if there are multiple spinning drives in the same enclosure, getting drives with vibration resistance is a good bonus. Most drives listed for NAS use will have this extra vibration resistance.

Yeah, you don’t want a surveillance drive. They are optimized for continuous writes, not random IO.

It’s probably worth familiarizing yourself with the difference between CMR and SMR drives.

If you expect this to keep growing, it might make sense to switch to SAS now - then you can find some really cheap enterprise class drives on ebay that will perform a bit better in this type of configuration. You’d just need a cheap HBA (like a 9211-8i) and a couple breakout cables. You can use SATA drives with a SAS HBA, but not the other way around.

Apart from the SMR vs. CMR, if your NAS will run 24/7 you need to make sure to use 24/7 capable drives or find a way to flash a 24/7-specific firmware/setting to a consumer drive. Normal consumer drives (e.g. WD Green) tend to have a lot of energy saving features, e.g. they park the drive heads after a few seconds of inactivity. This isn’t a problem with normal use as an external drive that only gets connected once in a while. But in a 24/7 NAS the drive will wake up lots of times and park again, wake up, park again … and these cycles kill the drive pretty fast.

https://www.truenas.com/community/threads/hacking-wd-greens-and-reds-with-wdidle3-exe.18171/

Lots of good advice here. I’ve got a bunch of older WD Reds still in service (from before the SMR BS). I’ve also had good luck shucking drives from external enclosures as well as decommissioned enterprise drives. If you go that route, depending on your enclosure or power supply in these scenarios you may run into issues with a live 3.3V SATA power pin causing drives to reboot. I’ve never had this issue on mine but it can be fixed with a little kapton tape or a modified SATA adapter. It’s definitely cheaper to shuck or get used enterprise for capacity! I’m running at least a dozen shucked drives right now and they’ve been great for my needs.

Also, if you start reaching the point of going beyond the ports available on your motherboard, do yourself a favor and get a quality HBA card flashed in IT mode to connect your drives. The cheapo 4 port cards I originally tried would have random dropouts in Unraid from time to time. Once I got a good HBA it’s been smooth sailing. It needs to be in IT mode to prevent hardware raid from kicking in so that Unraid can see the individual identifiers of the disks. You can flash it yourself or use an eBay seller like ThArtOfServer who will preflash them to IT mode.

Finally, be aware that expanding your array is a slippery slope. You start with 3 or 4 drives and next thing you know you have a rack and 15+ drive array.

The concern for the specific disk technology is usually around the use case. For example, surveillance drives you expect to be able to continuously write to 24/7 but not at crazy high speeds, maybe you can expect slow seek times or whatever. Gaming drives I would assume are disposable and just good value for storage size as you can just redownload your steam games. A NAS drive will be a little bit more expensive because it’s assumed to be for backups and data storage.

That said in all cases if you use them with proper redundancy like RAIDZ or RAID1 (bleh) it’s kind of whatever, you just replace them as they die. They’ll all do the same, just not with quite the same performance profile.

Things you can check are seek times / latency, throughput both on sequential and random access, and estimated lifespan.

I keep hearing good things about decomissioned HGST enterprise drives on eBay, they’re really cheap.

Other people have suggested good info to gain nuisanced knowledge. I recommend starting with a simple fact. With enough time and/or the right conditions all storage will fail. Design your setup with redundancy. I personally had to replace 2x 12tb drives this year. I have raidz3 (3 parity drives) and a hot spare. So I just bought cheap replacements from a reputable seller on eBay and consider it part of the cost of self hosting.

It depends what your parameters are. For spinning hard disks, you want to look at total power cycles, and mean time between failures. More enterprise drives have very long mean time between failures

In fact for spinnig hard disks, turning on can be on a likely failure mode, so there’s machines out there if you power off there’s a good chance they won’t come back on in the enterprise data centers

Your solid state hard disks, you want to look at meantime between failures, but also total write volume. Enterprise discs tend to have much much much much much much greater write capacity

So all of these trade-offs cost money, if you’re looking at archival, where you write the data only once, then you can go with a disk that has a low total write volume

Read about the specific features of the “WD RED” drives. There are some pretty good articles out there, and you are going to learn a whole lot reagarding your question.

I got a bunch of them in my private server. I didn’t know all these details when I bought them LOL, but they do a good job, reliable, silent, for 6 years and counting.

I highly recommend watching this guys videos on his analysis of the backblaze data https://www.youtube.com/watch?v=IgJ6YolLxYE&t=1

And a comparison of the difference WD drive colours, which might not be what you expect https://www.youtube.com/watch?v=QDyqNry_mDo&t=2

Also remember that your parity has to be more or equal to the biggest drive in your array. If you buy a 10 but don’t have another 10 you must use the 10 as parity.

Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:

[Thread #750 for this sub, first seen 15th May 2024, 23:25] [FAQ] [Full list] [Contact] [Source code]

Yes three are differences but you’re running a redundant array of independent disks in order not to care about those differences.

It can make quite a big difference. https://www.backblaze.com/blog/backblaze-drive-stats-for-q1-2024/

I’ve read advice against buying used storage unless you don’t mind being at more risk of losing the data in there.