TR Forums

So if any of you know me from previous posts you know I spend a lot of time with enterprise hardware and consumer hardware workstations, but one think I know practically NOTHING about is consumer hard drives.

I have finally run out of space on my home server using WD Red 3TB drives. I'm looking for a bunch of 6TB or 8TB drives that I can run three or four of to build a larger array. My problem is that I like how quiet the Reds are, but I don't like how unreliable WD are becoming. Newegg user reviews corroborate the Backblaze numbers that show WD Reds to be highly unreliable. Four years of data across 82,000 drives shows me that the 3TB Red drives and the 6TB Red drives I was looking to replace them with are the second and third least reliable drives Backblaze have ever used with about a 6% chance of drive failure per drive, per year. That means my array is statistically lucky to have not experienced a failure, and I was forced to replace the previous array of 2TB Greens (now called Blues) because of the awful head-parking issue that I didn't realise until after the damage was done.

Unfortunately, almost all the drives on the market except for WD reds or nasty SMR drives that don't fit my use case are 7200rpm and the few 7200rpm drives I have are noisy as hell. I have a Toshiba X300 that is the noisiest thing I've heard since the 90's. I have a couple of WD black 6TB drives for workstation datasets which are fast but too noisy, and I have a pre-WD HGST 4TB Ultrastar which is quieter than the X300 or blacks but still pretty noisy.

So, I'm looking for 6-8TB and I want to be unable to hear it at idle even when it's under the desk I'm sitting at.
Any gerbil recommendations or do I just have to find somewhere else in the house to move my new hard drives, because they're too noisy to be in the same room as?

I know that Backblaze is only one specific usage case and not necessarily a true indicator of desktop/server drive reliability, but in the absence of better data, and with star-ratings on the 'Egg with several hundred or even thousands of reviews, I'm seeing at least some similarity between Backblaze's numbers and the user reviews. I certainly won't be buying the nice, quiet 5400rpm WD Blues/Reds again.

Thanks in advance, y'all.

Modern HDDs are pretty quiet for the most part. They have been that way ever since they moved to fluid bearings. You get better luck with using noise damping kits/mounts if you feel that modern HDDs are too loud for your tastes.

Thanks but I only mentioned fluid bearing'ed models, so there's stil a huge difference in noise between vendors and models.

I've been looking for sites that give decibel readings for their tests but there aren't many sites that even bother with mechanical storage anymore, and the old staple (SPCR) hasn't updated their article since 2014, which is too long ago to cover any of the capacities I'm interested in.

The kits you mention are usually 5.25" > 3.5" bays, right - and either increase drive temperature or add a whiny little 40mm fan? I just thought I'd ask in case there's something else that I've overlooked.

2.5" drives.

I went with the "run a network cable to the crawlspace and put the file server there" approach years ago. The only downsides have been 1) since it is headless, on those rare occasions where it needs physical maintenance it has typically been untouched long enough that it is walled in with piles of junk which have to be cleared out of the way first; and 2) a full tower with hard drives stuffed into nearly every available bay is really heavy, hard to lift into the crawlspace, and hard to move around once it is in there. #1 is all on me and my packrat habits; I've partially solved #2 (the moving it around in the crawlspace part) by putting it on a cheap mover's dolly from Harbor Freight.

Krogoth wrote:

Modern HDDs are pretty quiet for the most part. They have been that way ever since they moved to fluid bearings. You get better luck with using noise damping kits/mounts if you feel that modern HDDs are too loud for your tastes.

IDK, I have a WD4003FZEX and it's loud enough to be annoying

Duct Tape Dude wrote:

2.5" drives.

Cost/GB is higher with that approach.

You can cram a lot of them into a small space though, with devices like this: https://www.newegg.com/Product/Product. ... -_-Product (unfortunately it has fans; but I imagine with a little negative case pressure to draw air in past the drives you could probably disable the fans without overheating the drives)

The other problem with 2.5" drives is that the capacity is too low. 2TB for standard 9.5mm drives, 4TB in a 15mm drive before they cheat with SMR and destroy the performance.

With 15mm ones you can't use those multi-bays but that doesn't matter because at 4TB I'm not really getting a capacity upgrade before I run out of SATA ports.

I can tell you, on the 2.5" side, a huge differentiator is the casing. There are the noisy ones that are basically tin cans you see with SSDs and there are super quite ones that are made out of block of aluminium and have heft to them to absorb/muffle arm movements, vibrations and sounds.

Also, are you aware of, "New Price Drop - 8TB WD External Easystore USB3 (8TB WD Red Inside)" and (same) "Harvesting WD easystore external drives for NAS drives?"

just brew it! wrote:

Duct Tape Dude wrote:

2.5" drives.

Cost/GB is higher with that approach.

I actually find high capacity 2.5" and 3.5" drives to be roughly on par with one another pricewise. You can find 5TB 2.5" for $130 ($26/TB) and 8TB 3.5" for $160 ($20/TB). It's not orders of magnitude off and you get higher density and lower power consumption. On average it takes about 5 2.5" drives to equal one 3.5" drive in terms of power consumption. They run cooler, less likely to require fans, etc.

Imo for quiet drives, you just can't beat a cluster of 2.5"s.

Chrispy_ wrote:

The other problem with 2.5" drives is that the capacity is too low. 2TB for standard 9.5mm drives, 4TB in a 15mm drive before they cheat with SMR and destroy the performance.

With 15mm ones you can't use those multi-bays but that doesn't matter because at 4TB I'm not really getting a capacity upgrade before I run out of SATA ports.

If your workload is random enough to be affected by SMR then that's definitely fair, but remember that many high capacity 3.5" drives do the same exact thing. For sequential writes I still get a sustained 120MB/s out of a shucked 4TB Seagate and it runs dead quiet compared the 5TB 3.5" sitting nearby.

I was forced to replace the previous array of 2TB Greens (now called Blues) because of the awful head-parking issue that I didn't realise until after the damage was done.

Silly question, but why don't you replace the Reds with newer Blues, and just disable (or extend the time of) the head parking? I did that on my 2TB greens immediately after buying them, and they've been doing wonderfully in my 3-drive FreeNAS box for a few years now.

I'm not sure if this still works on the current model drives, but this is the procedure that I used a few years ago to disable/lengthen the head parking time on my drives:
https://wdullaer.com/blog/2015/04/05/hack-your-wd-greens/

The 3TB HGST Deskstar NAS drives I have been using, and put in many more systems have had zero complaints. Honestly, I'm not sure i've ever even heard mine.

They have 4/6/8 TB versions as well.

https://www.newegg.com/Product/Product. ... 6822146116

Veerappan wrote:

I was forced to replace the previous array of 2TB Greens (now called Blues) because of the awful head-parking issue that I didn't realise until after the damage was done.

Silly question, but why don't you replace the Reds with newer Blues, and just disable (or extend the time of) the head parking? I did that on my 2TB greens immediately after buying them, and they've been doing wonderfully in my 3-drive FreeNAS box for a few years now.

I'm not sure if this still works on the current model drives, but this is the procedure that I used a few years ago to disable/lengthen the head parking time on my drives:
https://wdullaer.com/blog/2015/04/05/hack-your-wd-greens/

This has been a tempting option, and practically the only 5400rpm one that I can think of myself (and why I asked for gerbil advice!).

I know about disabling the head parking but my experience with the greens is that they were cheaper than the reds and therefore noisier and less carefully balanced. Given the statistically terrible reliability of the WD disks in general, I'm trying to not repeat the mistake of buying more. I'm also curious why WD still make models that head-park when it has a long history of causing problems in anything other than a single-drive desktop situation.

Hmm, I found something on Hardware.info from this year:

https://uk.hardware.info/reviews/7265/14/nas-hdd-review-18-models-compared-noise-levels

Looks like the WD Red is crazy quiet and other reviews say it's basically a rebranded HGST He8 drive design from before the WD buyout, just tweaked to run at 5400rpm.
That means it *should* have HGST build quality and reliability, rather than WD's lousy track record. More importantly, I know He8 drives are solid, I have a rack of them at work and they probably have more trouble-free operational hours between them than I'm likely to ever manage at home.

Helium is notoriously difficult to confine for extended periods of time. My concern with helium drives is that the far end of the bathtub curve will be really steep, as the helium drops below the levels required to sustain proper head flying height.

I know that anecdotal evidence isn't worth much, but I have four Reds in my array, three 5TB models about 8-10 months old, and a recently-introduced 6TB spinner. No problems so far (knock on wood).

They're inside an Antec P182 with rubber mounts and good cooling, so I'm fairly sure that helps a fair bit.

The only time I hear my 4 GB HGST (in a P182 as well) is at startup when a certain app needs to read a large amount of data in its init cycle and I can hear the seek chatter.

Chrispy_: Not sure if this helps, but the 8 or 10 TB WD Red or the 10 TB WD Red Pro are super quiet spec'd at 20 dB idle / 29 seek. Simply, nothing else but the WD Purple is that quiet. I too like this, and have used Red Pro's for NAS's in recent times, and I hope having triple-mirroring and client devices keeping copies of their data will render the reported unreliability of WD drives a non-issue.

If you really want to avoid WD, other options are (obviously) only Toshiba or Seagate nowadays. Personally, I'd stick with NAS (or maybe surveillance) drives for a NAS to avoid f/w hammering away on sector issues.

As you know, Toshiba drives are much noisier, and its high-capactity ones are only the NAS N300, spec'd as typical 33 dB idle / 35 dB in use, and the X300 is only other one even at 6 TB and is 34 dB at idle with no published noise level for in use. (That's even worse than the old Ultrastars.)

So then there's only Seagate, which I have no recent experience with. It's NAS drive IronWolf isn't spec'd as quiet as a WD but isn't rated too badly at typical 27 dB idle and 28 usage for the 6, 7, or one of the 8 TB models, and at 28 / 32 dB for the other 8 TB and the 10 TB one. I do like that the IronWolf drives have a decent UBER of 1 in 10¹⁵ too, the WD Reds and Toshiba N300's are both spec'd to 1 in 10¹⁴ which counts as shoddy in my book when the drives have 0.64 x 10¹⁴ bits ... The IronWolf Pro doesn't have an acoustic rating in its spec sheet so might well be noisy.

Backblazes' numbers are likely much higher than anything you'll see at home. They actually work the drives fairly hard compared to most NAS duties in a typical installation. UBER is pretty similar across all drives regardless of the spec (and well above it, generally).

I have a 6 TB Seagate Enterprise Capacity drive that's fairly quiet as well as a 10 TB Seagate Skyhawk that's even more quiet. Almost all drives are quiet these days compared to the clunkers of even a decade ago.

(pre-post admission: I have not read the rest of the thread)

I don't have a lot to add here because I don't care much about noise and don't pay attention to it, but I will note that the Toshiba DT01ACA300 (aka HDKPC08?) drives I've been using and recommending for a few years now are inaudible in operation. I have to literally press my ear to the drive to hear it working because I can't feel it by touching it.

Waco wrote:

Backblazes' numbers are likely much higher than anything you'll see at home. They actually work the drives fairly hard compared to most NAS duties in a typical installation. UBER is pretty similar across all drives regardless of the spec (and well above it, generally).

Ta. The old 2007 Google consumer disk study of a 100,000-strong selection of drives from different parts of their fleet concluded that there was no correlation between failure rates and how hard the disks were worked. Indentical sample sets from the most punished scenarios were failing at exactly the same rate as mostly-idle disks doing almost nothing. Likewise, on the condition that the temperature was within manufacturer tolerances, whether a drive was hot or cool also had no effect on failure over time.

For this reason (alone) I am happy to treat more recent failure-rate data from Backblaze as representative of consumer hard disk failure rates today. Sure, Backblaze's disks probably work harder than my disks and run hotter in dense enclosures than my little server, but the only thing that seems to matter is the quality of the drive in the first place and that comes down to manufacturer variance and which model you buy (Budget consumer, Home NAS, Prosumer, Enterprise nearline, Enterprise Tier-1 etc)

Waco wrote:

I have a 6 TB Seagate Enterprise Capacity drive that's fairly quiet as well as a 10 TB Seagate Skyhawk that's even more quiet. Almost all drives are quiet these days compared to the clunkers of even a decade ago.

Thanks (and thanks to others too). X is quieter than Y is extremely useful. It's not scientific but even if I don't know what X or Y sound like, if someone else says Y is quieter than Z and I own Z, we're golden. Hearing is all relative rather than absolute anyway!

Oh, in case it's not clear by now: WD Reds are by far the quietest "fast" drives I've ever laid ears on.

Chrispy_ wrote:

Waco wrote:

Backblazes' numbers are likely much higher than anything you'll see at home. They actually work the drives fairly hard compared to most NAS duties in a typical installation. UBER is pretty similar across all drives regardless of the spec (and well above it, generally).

Ta. The old 2007 Google consumer disk study of a 100,000-strong selection of drives from different parts of their fleet concluded that there was no correlation between failure rates and how hard the disks were worked. Indentical sample sets from the most punished scenarios were failing at exactly the same rate as mostly-idle disks doing almost nothing. Likewise, on the condition that the temperature was within manufacturer tolerances, whether a drive was hot or cool also had no effect on failure over time.

I would argue that that study is so outdated compared to modern drives that it doesn't apply. I absolutely see workload-related fallout at work, the sample size is on the order of many tens of thousands of drives over the course of the last half-decade.

There are definitely good and bad models too. The problem is, you typically don't know which models are problematic until it is too late.

Waco wrote:

UBER is pretty similar across all drives regardless of the spec (and well above it, generally).

UBER for all current drives is quoted as either 1 in 10¹⁵ or 1 in 10¹⁴, why would that be if the error rate on all drives was "pretty similar" to that of those with a spec'd error rate an order of magnitude lower?

Topinio wrote:

Waco wrote:

UBER is pretty similar across all drives regardless of the spec (and well above it, generally).

UBER for all current drives is quoted as either 1 in 10¹⁵ or 1 in 10¹⁴, why would that be if the error rate on all drives was "pretty similar" to that of those with a spec'd error rate an order of magnitude lower?

I'm just giving some insight into the drive population I have. Anecdotes are anecdotes, but I've seen many tens of thousands of them and UBER ratings have never been close to what we see in the field.

Chrispy_ wrote:

The old 2007 Google consumer disk study of a 100,000-strong selection of drives from different parts of their fleet concluded that there was no correlation between failure rates and how hard the disks were worked. Indentical sample sets from the most punished scenarios were failing at exactly the same rate as mostly-idle disks doing almost nothing..

Kind'of like wine. Quality (longevity) mostly down to when/where it was produced and how it was stored/transported.

(I also like 2.5" drives as far as noise goes, but if you need drive-space and are running-out of sata ports then it's a poor solution.)

I would say though that you should always wait for some sort of sale and purchase in relative "bulk" - which is what I did with my 2.5" drives (..just 5 drives of 750GB at a little under $30 including shipping for each on sale on Amazon.. plus some "cash-back" from the credit card.)

https://slickdeals.net/deals/3.5-inch/
https://slickdeals.net/deals/drives/

morphine wrote:

Oh, in case it's not clear by now: WD Reds are by far the quietest "fast" drives I've ever laid ears on.

Define "fast"

My Reds are fast enough for me, but they're bottom of the pile in every benchmark ever, slowest drive on test, "worst transfer rates we've seen in years" sort of performance. I guess that's what you get when you reduce the spindle speed and modify the actuator speed to prioritise low vibration and power consumption! It's a choice I'd gladly make again given that most of the time I'm accessing the disks over a Gigabit ethernet anyway.

My 3TB Reds are (individually) something like 130MB/s at best and 15-20ms access times, which is terrible by even the typical drives of five years ago, but these days most drives are in the 200MB/s ballpark and closer to 10ms access time. Still, with a 110MB/s ethernet bottleneck, sequential performance of a single disk just isn't that important. They'd be throttled by the Gigabit connection at just 40MB/s each....

"Fast" as in way better than most consumer SSDs I've used, even considering the 5400 RPM limitation. Even though they don't have insane STR, I've found that the NCQ implementation actually works. Copy two things over at the same time and the speed drops to roughly half instead of just stopping dead.

I used "fast" because we're still talking 5400 RPM drives (WD Red Pros are the 7200 RPM). For mass storage purposes, they're plenty nice. If someone's going to use these as system drives... well, it's their funeral.

I think the quietest high capacity drive on the market would be the Seagate 5TB Backup Plus Portable. It's 2.5", slow spindle speed, and enclosed in a case that should keep sound to a minimum. Not quite up to the 6-8TB from your original post, but if you're really picky about noise it's an option.