12

Is it a good idea to use a hard drive that was made for use in NAS as a desktop system drive?

I am particularly concerned about error recovery.

The following post claims that NAS drives do not try as hard to fix disk errors, compared to normal desktop drives:

https://www.reddit.com/r/DataHoarder/comments/78ek5h/comment/doxiq91/

BarraCuda drives are engineered for use in a desktop environment, whereas NAS drives like the IronWolf are designed for use in NAS enclosures. IronWolf drives are engineered to excel in situations where they work in a NAS as RAID teams, meaning they don't as aggressively try to fix reallocated/bad sectors because they can just "move on" and leave something for another drive in the array, meaning when used as a single drive in a desktop, it may potentially result in these errors stacking up quicker than a standard desktop grade drive would.

Is there any truth to this?

I would like to use a Seagate IronWolf drive as a desktop drive, without RAID.

Improve this question
4
  • 5
    I have used an IronWolf drive for years in my desktop without an issue.
    – Ramhound
    Commented Nov 29 at 18:45
  • 3
    It's possible, they're still HDDs - if buying one, you may pay more for features you're not going to be using since NAS drives are optimized for RAID, whether it be software or hardware based. (Anecdotal: I've seen nothing but issues running Seagate's IronWolf NAS drives on TrueNAS servers [previously FreeNAS], with minimal issues using WD's Red NAS drives - mainly that IronWolf drives fail prematurely.)
    – JW0914
    Commented Nov 29 at 21:00
  • 2
    Basically I want to pay extra to get a CMR drive, I hate SMR. CMR desktop drives I have seen are even more expensive than NAS drives, at least at >1TB capacities.
    – mcu
    Commented Nov 29 at 21:36
  • @mcu Something else to keep in mind are 7200RPM will often require buying the Pro version of NAS drives, albeit WD's Black PC gaming drives only come in 7200RPM versions, which are the same price as the Red Pro drives. (Anecdotal: I run 8x Red Pro drives in a SilverStone DS380 desktop server case and heat dissipation from the drives has been an issue in the past, so depending on the bay layout of the case in relation to fans and other high temp components, it's something that may want to be factored in)
    – JW0914
    Commented Nov 30 at 11:44

3 Answers 3

Reset to default
15

IronWolf drives are engineered to excel in situations where they work in a NAS as RAID teams, meaning they don't as aggressively try to fix reallocated/bad sectors because they can just "move on" and leave something for another drive in the array

This sounds like the drive's firmware just has a different retry timer before it gives up on a failing 'read' command.

Probably all HDDs will automatically retry a failed read. The OS issues a READ command just once and waits, and if the HDD detects that it's unable to read the requested sector 100% correctly, it'll automatically try again, and again, and again, until it either succeeds, or gives up and responds with ERROR maybe a minute later.

But "a minute later" is much too long in situations where the disk is part of a RAID array, and the OS (or the RAID controller) would much prefer to detect the failure quickly because it can easily obtain the same data from a mirror or recalculate it from parity.

So as far as I know, it's very common for "NAS optimized" drives to have a different firmware variant that significantly reduces the drive's automatic retry timeout in order to let the RAID do its work instead.

Improve this answer
3
  • The thing about NAS rotating head disks is you want to turn off sector remapping and let the higher level protocol handle it. The performance degradation of sector remapping on a RAID array is too great and it's better to let the stack on top of it remap the block for all the disks.
    – Joshua
    Commented Dec 2 at 19:17
  • So, if a NAS drive stops error recovery after 7 seconds, but a desktop drive after 60 seconds (may not be actual numbers), how likely is it that the error would get fixed after 7 seconds but before 60 seconds? Are there are any statistics on this?
    – mcu
    Commented 2 days ago
  • I have no idea. It's probably above zero, though, given that I've heard HDD storage at modern-day densities being described as "practically radio signal processing" (something along those lines) and a matter of picking up enough signal vs noise that the built-in error correction could turn it into data, in which case it's possible that a valid sector could be recovered from multiple reads (like how phone cameras do "HDR")... but without knowing much anything about the proprietary algorithms that HDD firmwares use, this is only wild guessing.
    – grawity
    Commented 2 days ago
8

For me the things are overexposed. What I mean is the behavior of "don't as aggressively try to fix" is true but the difference in this direction between NAS and desktop versions is not so big. The above is true mainly for surveillance disks because for this role in required to have as much as possible constant flow of information from cameras to the disk(s). And small error/pixel(s) degradation do not ruin the record.

So my opinion is you can use NAS disk(s) but you should be aware they emit more heat and they are a bit noisier (compared to desktop models)

Improve this answer
3

This comes down to the error-recovery behavior of the drive. In general:

  • Consumer drives (including Seagate’s Barracuda/Firecuda models and WD’s Blue/Green/Black models) are designed under the assumption that the only option for recovering data from a bad sector is for the drive to do it itself because most end-user systems don’t do RAID and most home-users historically didn’t do proper backups (if they did any at all). This leads to them taking a long time (sometimes multiple minutes) trying to re-read a bad sector in an attempt to get data out of it.
  • Enterprise drives (including Seagate’s Exos models and WD’s Gold models) are designed instead under the assumption that not only are there other options for recovering data from bad sectors, but that those are the preferred options in most cases. As such, they tend to give up relatively quickly (I believe 7 seconds or so is the norm) so that the error gets reported promptly up the storage stack and whatever other recovery mechanism is in place gets triggered, but they also have support for configuring the error recovery behavior.
  • NAS drives (including Seagate’s IronWolf models and WD’s Red models) are a kind of in-between case, but they provide similar error recovery behavior to enterprise drives. Rather importantly, they also usually support configuration of this behavior, though actually configuring it generally requires using something other than Windows.

Rather importantly, you can almost always configure error recovery behavior for a NAS drive to be like a desktop drive.

However, I would also expect the spindle motor on a NAS drive to not hold up as well in typical desktop usage. NAS drives are usually designed for 24/7 operation, but the normal use-case for a desktop system involves the drive starting and stopping frequently (either because of power management (spinning down the platters to save power, or because of the system being powered off frequently), and that type of usage generally prefers different types of bearings and different motor design than continuous operation with infrequent starts.

Improve this answer
3
  • 2
    Rather importantly, you can almost always configure error recovery behavior for a NAS drive to be like a desktop drive. - How? Jumpers? Standard SATA commands? Vendor-specific program to send special commands? I don't remember seeing a mention of this in Linux HD utilities like smartctl or hdparm.
    – Peter Cordes
    Commented Nov 30 at 19:43
  • 3
    @PeterCordes For SATA devices it’s the SCT ERC commands, which can be invoked via smartctl. I believe there’s some other command set for SCSI.
    – Austin Hemmelgarn
    Commented Dec 1 at 2:10
  • 2
    > NAS drives are usually designed for 24/7 operation: i'm not sure this is entirely accurate (or at least, i'm not sure it ought to be). many consumer/prosumer NAS solutions such as all desk-top synology boxes actually do power management of the drives as well, though they aren't as agressive as a PC.
    – Themoonisacheese
    Commented Dec 2 at 9:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .