Highest Capacity Hard Drive with no SMR?


26 posts in this topic Last Reply

Recommended Posts

Fellow Forum Members,

I do not trust hard drives that use Shingled magnetic recording (SMR) technology to store my data.  So I called Western Digital tech support so he can clarify for me which hard drive capacities I need to stay away from to avoid SMR.  I was surprised when he told me none of their  8TB, 10TB, 12TB or 14TB hard drives use SMR.  Is the WD tech speaking the truth? In short, what is the highest capacity Western Digial hard drive I could buy that uses traditional magnetic recording to the platter and does not rely on using SMR or any other form of exotic data stacking recording technology?  12TB hard drives is the capacity I am interested in so am I correct to believe this capacity utilizes  traditional magnetic recording to platter technology? 

 

Lastly, I need to buy 5 hard drives for the Unraid system I am building. However,  I am undecided on whether I should buy 12TB WD Red Pro hard drives or go with 12TB WD Gold Label Enterprise Sata hard drives.  Prices for the two are close together. So I would appreciate any opinions relating to which of the two is the most reliable?  Thanks in advance for the help. 

 

 

Link to post
I was surprised when he told me none of their  8TB, 10TB, 12TB or 14TB hard drives use SMR.  Is the WD tech speaking the truth?

Yes, currently WD only makes 2TB and 6TB 3.5" SMR drives.

 

Note that all current 3.5" Toshibas drives up 14TB are CMR, and AFAIK all Seagate drives from 10 to 16TB are also CMR, Seagate does have an 8TB SMR.

 

 

Link to post
7 hours ago, binar said:

Lastly, I need to buy 5 hard drives for the Unraid system I am building. However,  I am undecided on whether I should buy 12TB WD Red Pro hard drives or go with 12TB WD Gold Label Enterprise Sata hard drives.  Prices for the two are close together. So I would appreciate any opinions relating to which of the two is the most reliable?

Good luck on this one.  I am going to refer you to this blog from Backblaze as they are the only ones who have ever posted up any data on drive failure rates.

 

      https://www.backblaze.com/blog/backblaze-hard-drive-stats-q3-2019/

 

As you can see from this, the failure rates of all drives are very low.  There are variations between models from the same manufacturer.   While I have not analyzed the data that closely for this quarter, it often seems that this model variation may be larger than between manufacturers. 

 

Now you also have to realize that Backblaze usage pattern is not exactly like the one of the typical Unraid user but it is the only data that is available in the 'wild' for large numbers of drives.  I believe that their purchase decision is primarily made on pricing.  (They are looking for the lowest cost per TB!)  That could be why there are fewer WDC drives in the recent reports.  (You would have to read the older quarterly reports to find if this supposition is true.)   That is also why you don't find any 'Enterprise' class drives in their data.  (I believe they think as I do that all hard drives have basically the same quality and only difference is the marketing strategy-- pricing vs warranty costs.   I.e., A longer warranty implies higher quality to the buyer but in reality, the increased price will cover the expected warranty returns!)

 

I will make a couple of more observations.  First, every hard drive manufacturer will have manufacturing lots of drives that have lower quality levels.  Therefore, you can protected yourself from buying into one of these lots by not buying a larger quantity of drives at the same time from the same source.  Concealed shipping damage is another issue.  We all know that packing practices varying between vendors (and even between individual packers at the same vendor) and the physical abuse that the package receives varies with each carton, so buying from different sources at different times may be wise. 

Link to post

@Frank1940

There are grains of truth in what you wrote.  An article/interview I recently read from backblaze

https://www.backblaze.com/blog/how-backblaze-buys-hard-drives/

lays out a lot of your points.

 

For the TL;DR crowd...

 

Quote

 

A final question. In your opinion, are hard drives getting better?

Yes. Drives are amazingly durable for how hard they’re used. Just think of the forces inside a hard drive, how hard they spin, and how much engineering it takes to write and read the data on the platters. I came from a background in precision optics, which requires incredibly precise tolerances, and was shocked to learn that hard drives are designed in an equally precise tolerance range, yet are made in the millions and sold as a commodity. Despite all that, they have only about a 2% annual failure rate in our centers. That’s pretty good, I think.

 

 

 

 

We haven't had a WD Caviar or Seagate Click-Of-Death fiasco in a long time; those were in the 2000s.

(although to this day I still can't forgive Seagate for that, or the 90% failure rate (not a joke!) I had on one of their enterprise differential SCSI drive lines. That was a dark time persoanally.)

Edited by sota
Link to post

Thank you to  all for your posts. I have learned there are currently three type of recording technologies which I have listed below:

  • Conventional Magnetic Recording (CMR)  (aka Longitudinal Magnetic Recording (LMR)).
  • Perpindicular Magnetic Recording (PMR)
  • Shingled Magnetic Recording (SMR)

I called Western Digital again and this time the tech told me he could not disclose which of the three recording technologies their 12TB an 14TB Gold Label server grade hard drives use.  So I'm back to square one.

 

Can anyone out there please point me to a website that will clarify for me what recording technologies do 12tb and 14tb hard drives use categorized by Hard Drive Manufacturer.   What I have noticed is hard drive manufacturers do not really want to disclose this information and I believe as a consumer I have the right to know such details.  Any help will be greatly appreciated. 

 

Link to post

Johnnie Black,

Thank you for your post. From what I have read SMR hard drives have a write speed penalty associated with them. This is why I want to stay away from SMR drives.   I found what you say about PMR and SMR being the same very interesting. Because after visualizing it in my head in 3D I can understand why technically PMR and SMR are the same.  To confirm, in your opinion the two hard drives options I have listed below use CMR:

 

$343 - Seagate Exos X14 14TB (ST14000NM0018): https://www.newegg.com/p/1Z4-002P-00RY9

$490 - WD Gold 14TB Enterprise Class Hard Drive (WD141KRYZ):   https://www.newegg.com/p/N82E16822234400

 

And if I understand you correctly, CMR is really PMR. Honestly, I find this confusing. I guess the best question to ask is as follows. Does CMR have a write penalty associated with it like SMR does?

 

Lastly, there is a $147 price difference between the two hard drive options I'm interested in buying shown above. Does such a big price difference necessarily mean the Seagate drive is designed not to last as long compared to the Western Digital Gold 14TB hard drive?  Any opinions welcome.  Thanks in advance.

 

Link to post

Yes, both are CMR, no write penalty, and SMR write penalty depends mostly on what they are used for, for example they work great with Unraid and usually no different than CMR drives, they work very bad with any kind of RAID, like ZFS, BTRS, hardware raid, etc.

 

For example I use a four disk btrfs raid5 pool with SMR drives, 2TB 9.5mm 2.5" drives currently only exist with SMR, and write penalty is very obvious, though OK for me for this case, just want small low power drives, write performance is not important for this since it's a temporary archive.

 

Write speed starts at around 300MB/s which is about what I expected, but after about 30 or 40GB transferred (maybe they have a small CMR zone?) it tanks to around 30/50MB/s and stays like that for the duration.

 

 

image.png.4d9a1c91a2652e00066e3fa160489ca5.png

 

 

 

Link to post
32 minutes ago, Spies said:

What's the comparative speed when writing to a relatively full SMR disk in unraid?

Like mentioned with Unraid SMR usually performs the same as CMR, at least in my experience, I got close to 200MB/s sustained write speed when writing to the outer cylinders of 8TB Seagate Archive drives.

Link to post
5 hours ago, Spies said:

What's the comparative speed when writing to a relatively full SMR disk in unraid?

It depends on the size of the write and how much of it is sequential (vs random).

 

For highly sequential writes, my (old) 8TB Seagate Archive behaves just like typical 5400rpm drives e.g. with 7+TB filled can easily take 100GB dump in 1 go at 100+MB/s.

(note, the SA is very fast early, comparable to WD Black, but slow down to WD Red 5400rpm level towards the end, hence, 100+MB/s)

 

For highly random writes, it gets down to less than 30MB/s after about 30GB or so i.e. way slower than a typical 5400rpm drive after the CMR "cache" fills up.

 

One thing you have to pay attention to with SMR drives (in my own experience, the Seagate Archive) is that it is not designed for heavy random write (which is NOT typical Unraid usage). I managed to kill a Seagate Archive by writing so much random data to it that (I think) the CMR cache died. It was during warranty so I got a replacement without any fuss.

Link to post
  • 1 month later...

As a rule of thumb any drive with 1.6TB platters or less will be CMR. 2.0TB = SMR. Also, cheaper $/TB drives tend to be SMR since they require fewer platters. Google around will find how many platters are in various drive models. I use a mix of 10TB (I just bought what was cheapest at the time I needed it), both CMR with 7 platters and SMR with 5 platters. Both are fine for my use (fill the drive once and then never rewrite it for archiving and only use it for reading on occasion). SMR does suck if you fill it and then have to delete/rewrite parts of it in typical PC use (the rewrite performance is horrendous, less than 1/10 the rate of a fresh format). As far as reliability, how important is your data? if it's irreplaceable don't rely on reputation or warranty and assume your particular drive will last forever, 0.33% failure rate is not 0%. Have backups (I have 3 copies; HDD, tape, and cloud, so even if the house burns down I still have the data, unless civilization ends and takes the internet with it as well before I have a chance to re-download it all from cloud backup).

Link to post
  • 2 months later...

It does not matter how much existing data you have on a "device managed" SMR drive. What matters if how many GBs you write per hour. If you write less than ~20 GB per hour per drive you'll be fine. If you write faster, the performance will be around 1-5 % of the usual until you (or bad latency) slow down enough for the drive to flush the PMR/CMR cache area. For short enough bursts, SMR drives are just fine. And they are fine for reading in all cases.

 

See also: https://unix.stackexchange.com/a/489530/20336

Link to post
9 hours ago, mtrantalainen said:

What matters if how many GBs you write per hour. If you write less than ~20 GB per hour per drive you'll be fine. If you write faster,

Not necessarily, if you do sustained random writes then yes, you'll hit the SMR wall after a few GBs, for sequential writes you can write TBs without slowing down, I did that several times with some of my SMR disks.

Link to post
On 12/16/2019 at 2:16 AM, johnnie.black said:

CMR is now used instead of PMR since technically SMR is also PMR (with SMR), so to avoid confusion drives now are called CMR or SMR, 12 and 14TB WD Gold are CMR.

Actually, that is incorrect. CMR and PMR are technically the same.

"PMR, also known as conventional magnetic recording (CMR), works by aligning the poles of the magnetic elements, which represent bits of data, perpendicularly to the surface of the disk. Magnetic tracks are written side-by-side, without overlapping."

Link to post
1 hour ago, johnnie.black said:

They are, and SMR is also PMR (with SMR), that 's why now CMR is preferred when referring to conventional PMR drives, for example:

 

https://blog.westerndigital.com/wd-red-nas-drives/

Only problem is that recently WD has been sending out SMR drives as WD Red NAS drives, which can get ugly fast.

https://blocksandfiles.com/2020/04/23/western-digital-blog-wd-red-nas-smr-drives-overuse/

Link to post
33 minutes ago, Simba7 said:

Only problem is that recently WD has been sending out SMR drives as WD Red NAS drives,

That's old news, and the WD blog I linked above is also about that, WD clarifying which models are CMR and what which are SMR.

Link to post
  • 7 months later...
On 12/17/2019 at 11:34 AM, JorgeB said:

Like mentioned with Unraid SMR usually performs the same as CMR, at least in my experience

Sorry for buttin' in: I know that Unraid has no typical RAID write pattern, but what about disk replacements. If a CMR array disk becomes replaced by a SMR disk, the SMR disk becomes written to completely with several TBs. What about write penalty in this specific case? Will the resync last for a multiple of the usual timeframe then?

 

Link to post

In my experience some SMR disks work fine with Unraid, e.g., the Seagate drives I've have, both the 8TB Archive model as well as 4TB 2.5" drives, they are fast during a rebuild/sync and don't notice any big slowdowns during large sequential transfers (mostly large files, might be different with small files), on the other hand WD and Toshiba SMR drives behave normally during rebuild/sync but they get extremely slow during large transfers, even for large files.

Link to post

Here's an example of the slower SMR performance in some drives, due to a config change in one of my backup servers I need to re-transfer all the data, when doing this I disable parity so I can transfer 2 or 3 disks at the same time, using 10GbE (due to the SSH encryption this CPU can only handle about 300MB/s), all 3 disks started transferring within minutes of each other and the 2 SMR drives are going noticeable slower, and the difference will probably be larger for the remaining two disks since there won't be such a CPU bottleneck.

 

imagem.thumb.png.7259adb458102fa566ffb16247d66880.png

Link to post
  • 2 months later...

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.