The three major HDD manufacturers are selling slower drives, without telling us

[hitardo]

The three major HDD manufacturer - Western Digital, Seagate, and Toshiba - have been implementing slower technology (SMR) in their most recent revisions of their hard drives, without telling customers.

 

Originally, this was noticed by a user, who contacted BlocksAndFiles.com for further investigation.

This was felt as users inserting a new HDD drive to their RAIDs, but the new HDD (with SMR) kept pulling out of the RAID.

 

In my opinion, this is unacceptable, because manufacturers decreased the performance significantly of their products, while keeping the same marketing name.

You could watch today a video, read a column, or see the performance of a given HDD at your friend's house; and when you purchase the exact same drive (e.g. WD Red 4TB), you would be getting an inferior drive.

Even if you bought a drive in the past, and now hit the "Reorder this item" button, to increase your RAID, you would have issues.

Thus, I consider this a very bad precedent!

 

Sources

 

The news I read originally was from NotebookCheck.net, here:

https://www.notebookcheck.net/Western-Digital-Seagate-and-Toshiba-have-been-using-slower-technology-in-some-recent-hard-drives-without-telling-customers.461822.0.html

 

The original investigation was from BlocksAndFiles.com, which separated their articles by manufacturer:

Western Digital: https://blocksandfiles.com/2020/04/14/wd-red-nas-drives-shingled-magnetic-recording/

Seagate: https://blocksandfiles.com/2020/04/15/seagate-2-4-and-8tb-barracuda-and-desktop-hdd-smr/

Toshiba: https://blocksandfiles.com/2020/04/16/toshiba-desktop-disk-drives-undocumented-shingle-magnetic-recording/

 

Additional sources:

https://www.tomshardware.com/news/wd-fesses-up-some-red-hdds-use-slow-smr-tech

 

Quotes

NotebookCheck.com:

Quote

A series of articles by Chris Mellor from the storage specialist website, Blocks and Files, has led to discussions about Western Digital (WD), Seagate, and Toshiba using shingled magnetic recording (SMR) across various consumer drives.

 

 

Quote

SMR is a technology that allows manufacturers to fit more data on to a drive platter by overlapping the sectors while still leaving a clear area for the read head to pass over. This results in a drive that can read at regular hard drive speed but is much slower during writing due to the way it needs to handle the overlapping data.

Quote

The issue is that these SMR drives sit within product lines where the poor write performance could impact on the systems in which they reside. Without specifying that these drives are SMR, users aren't able to avoid them if their use case is one that would be impacted.

 

 

 BlocksAndFiles.com's Western Digital column:

Quote

 

Alan Brown, a network manager at UCL Mullard Space Science Laboratory, the UK’s largest university-based space research group, told us about his problems adding a new WD Red NAS drive to a RAID array at his home. Although it was sold as a RAID drive, the device 'keep getting kicked out of RAID arrays due to errors during resilvering', he said.

 

Quote

There is a similar problem mentioned on a Synology Forum where a user added 6TB WD Red [WD60EFAX] drive to a RAID setup using three WD Red 6TB drives [WD60EFRX] in SHR1 mode. He added a fourth drive to convert to SHR2 but conversion took two days and did not complete.

 

BlocksAndFiles.com's Seagate column:

 

Quote

 

 Some Seagate Barracuda Compute and Desktop disk drives use shingled magnetic recording (SMR) technology which can exhibit slow data write speeds. But Seagate documentation does not spell this out.

 

Quote

Geizhals, a German-language price comparison website, lists seven Seagate SMR drives. 

- Barracuda 2TB – 7,200rpm – SATA 6gig – model name – ST2000DM008

- Barracuda 4TB – 5,400rpm – SATA 6gig – ST4000DM004

- Barracuda 8TB – 5,400rpm – SATA 6gig – ST8000DM004

- Desktop HDD 5TB – 5,900rpm – SATA 6gig – ST5000DM000

- Exos 8TB – 5900rpm – SATA 6gig – ST8000AS0003

- Archive v2 6TB – 5,900rpm – SATA 6gig – ST6000AS0002

- Archive v2 8TB – 5,900rpm – SATA 6gig – ST8000AS0002

 

BlocksAndFiles.com's Toshiba column:

Quote

Toshiba told Blocks & Files yesterday that its P300 desktop disk drives use shingled magnetic recording technology (SMR), which can exhibit slow data write speeds. However, the company does not mention this in end user drive documentation.

Quote

According to the Geizhals price comparison website, Toshiba’s P300 desktop 4TB (HDWD240UZSVA) and 6TB (HDWD260UZSVA) SATA disk drives use SMR.

 

CHANGE LOG:

18/04/2020 @ 03:23pm GMT - Original post

19/04/2020 @ Unknown - Moderator changes (thank you!)

20/04/2020 @ 08:59am GMT - Added Tom's Hardware source, separated the Seagate quotes, and added a clearer introduction.

 

[williamcll]

Fortunately I can afford more SSDs.

[hitardo]

23 minutes ago, williamcll said:

Fortunately I can afford more SSDs.

Fortunately, I can also afford SSDs on my computer.

 

But, on my NAS, where I have my backups, photos, videos, and much more... HDDs is still the only way to go.

How much would I have spent on SSD to reach my 8TB of total capacity?

 

Now, I am looking to upgrade my NAS to a 4-bay one, with 3x 8TB HDD, and 1x 512GB SSD for caching.

This situation concerns me.

 

And, in fact, concerns all of us.

Because it is an OEM changing the specifications of a drive, lowering its performance, without letting the customer know.

You could be watching a performance graph for a WD Red 2TB, but when your new one arrives at your door, we are surprised with a slower performance.

And, all this negative performance, while you are paying the same as before.

 

Is this fair?

Is this Legal?

Is this acceptable?

 

Especially when all of these OEMs have expensive drives, reputable names, and loyal customers.

This will be interesting to pay attention.

 

I, personally, will contact my local authorities and file a complaint!

[RejZoR]

I knew my 8TB Barracuda 5400 RPM was SMR drive and it's listed above. Not sure where I read it tho. They in fact don't state it anywhere on webpage or in their spec sheets. Probably on Geizhals since I use that page to find stuff in Europe.

 

I didn't care, because it's an archive drive. The hot stuff is on 2TB Samsung 850 Pro. But they really should be mentioning things like this. This stuff probably wouldn't ever surface if people using multiple drives in RAID arrays haven't noticed problems when they had regular and SMR drives inside same array.

[hitardo]

1 minute ago, RejZoR said:

I knew my 8TB Barracuda 5400 RPM was SMR drive and it's listed above. Not sure where I read it tho. They in fact don't state it anywhere on webpage or in their spec sheets. Probably on Geizhals since I use that page to find stuff in Europe.

 

I didn't care, because it's an archive drive. The hot stuff is on 2TB Samsung 850 Pro. But they really should be mentioning things like this. This stuff probably wouldn't ever surface if people using multiple drives in RAID arrays haven't noticed problems when they had regular and SMR drives inside same array.

Thanks for the tip.

Will start using it.

 

Yes, it depends on your application.

For archiving, yes, you will only feel when you are storing.

Then, for long term storage, there is no significant pani.

 

But, for NAS application - like WD Red or Seagate Barracuda - I think this is not acceptable.

 

Personally, I think this is a movement for OEMs to differentiate their Pro versions from their non-Pro versions.

This way, most users (especially SOHO and enthusiast), will spend more on their Pro lineup.

[wanderingfool2]

 

12 minutes ago, hitardo said:

But, for NAS application - like WD Red or Seagate Barracuda - I think this is not acceptable.

The Red drives are unacceptable.  They were advertised as NAS drives; so not putting in the use of technology that affects performance on read is very bad in my opinion.

 

The Barracuda drives though, I think, are more acceptable.  They are consumer grade drives, which are not rated for systems like NAS's...now if this was done on the IronWolf drives; then yes I would say raise your pitchforks.

[Zodiark1593]

31 minutes ago, wanderingfool2 said:

 

The Red drives are unacceptable.  They were advertised as NAS drives; so not putting in the use of technology that affects performance on read is very bad in my opinion.

 

The Barracuda drives though, I think, are more acceptable.  They are consumer grade drives, which are not rated for systems like NAS's...now if this was done on the IronWolf drives; then yes I would say raise your pitchforks.

Barracuda drives are more likely to be used as boot drives. SMR would be entirely unacceptable in this application as well. 

[Letgomyleghoe]

@seagate_surfer 

[Kilrah]

My 2.5" ST5000LM000 Seagates are also SMR - while not surprising in this particular case there is also nothing specifying it i the documentation nor any spec that allows finding it out before purchase.

And on the moderate capacity 4-8TB desktop drives mentioned above it really seems mean. I understand in the ultra high capacity models since well there is basically no other reasonable technical way to get those capacities, but on the standard stuff, especially on models which have been conventional for a long while there is absolutely no reason for the non-disclosed downgrade.

 

I really hope manufacturers don't get away with that. That should be class action lawsuit material.

[hitardo]

Guys, I understand your point of view.

But this case is more than that.

 

We are talking of and OEM changing the core of the product, without communicating the change.

 

Same as, a new phone:

Samsung Quiet S, with a Snapdrogn SoC.

Then, after selling millions of those, Samsung changes to:

Samsung Quiet S, with a MediaTek SoC.

 

What would you think of that?

Maybe call it Samsung Quiet S1, or S less.

But never call it the same.

 

Especially when the performance has decreased.

[Unixsystem]

5 hours ago, wanderingfool2 said:

 

The Barracuda drives though, I think, are more acceptable.  They are consumer grade drives, which are not rated for systems like NAS's...now if this was done on the IronWolf drives; then yes I would say raise your pitchforks.

I don't think it matters if they are Barracudas or if Seagate comes out with a new line called "Garbage Drives For Garbage People," if the drives used the be universally one standard of performance and then it gets changed to a grabbag of maybe worse performance, there needs to be either a rebrand or a concerted effort to inform consumers both on their spec sheet as well as on the various store listing what drive is running what technology. 

[Mihle]

All of these are bad, but WD Reds is the worst offender because Reds are more expensive.

 

It should be illegal to not spesify if it isn't already.

[mr moose]

5 hours ago, hitardo said:

Guys, I understand your point of view.

But this case is more than that.

 

We are talking of and OEM changing the core of the product, without communicating the change.

 

Same as, a new phone:

Samsung Quiet S, with a Snapdrogn SoC.

Then, after selling millions of those, Samsung changes to:

Samsung Quiet S, with a MediaTek SoC.

 

What would you think of that?

Maybe call it Samsung Quiet S1, or S less.

But never call it the same.

 

Especially when the performance has decreased.

 

I'm not sure I see a problem here,  Every time I buy a new hard drive (about 1 year on average) I don't assume the one I am getting is identical to one I bought a year or 3 ago.  Technologies and manufacturing processes change.  if you buy an i7 you wouldn't assume it was the same as last gens i7 would you?

[Unixsystem]

10 minutes ago, mr moose said:

 

I'm not sure I see a problem here,  Every time I buy a new hard drive (about 1 year on average) I don't assume the one I am getting is identical to one I bought a year or 3 ago.  Technologies and manufacturing processes change.  if you buy an i7 you wouldn't assume it was the same as last gens i7 would you?

It's a lot easier to find data and reviews about a specific CPU SKU than it is for an HDD. There's also the issue that most store pages online don't put the model number front and center, it's usually buried in the stats and a lot of times they will update the store page to the newest model without making it clear that anything has changed. If you buy a WD Red today on Amazon and then three years from now hit "Order Again" you will almost certainly wind up with a different model number, which in this case may be a totally different base technology. 

 

If I ordered an i7 today and then a year from now order another from them same tier (9700k to 10700k, for example) I would absolutely expect the new CPU to be at minimum the same performance with the same or improved technologies under the hood. If Intel put out an i7 tomorrow that had no AVX or one less core than the previous gen at the same price tier, that would be shitty. If they did that same thing but put it out as a revision of an existing SKU and didn't very clearly specify what changed, that would be bordering on criminal. 

[mr moose]

1 minute ago, Unixsystem said:

It's a lot easier to find data and reviews about a specific CPU SKU than it is for an HDD. There's also the issue that most store pages online don't put the model number front and center, it's usually buried in the stats and a lot of times they will update the store page to the newest model without making it clear that anything has changed. If you buy a WD Red today on Amazon and then three years from now hit "Order Again" you will almost certainly wind up with a different model number, which in this case may be a totally different base technology. 

 

If I ordered an i7 today and then a year from now order another from them same tier (9700k to 10700k, for example) I would absolutely expect the new CPU to be at minimum the same performance with the same or improved technologies under the hood. If Intel put out an i7 tomorrow that had no AVX or one less core than the previous gen at the same price tier, that would be shitty. If they did that same thing but put it out as a revision of an existing SKU and didn't very clearly specify what changed, that would be bordering on criminal. 

Go the products page for specs then.   just like other products with standard model numbers,  the tech changes everytime. 

 

You can check the advertised speeds and spec pretty easy:

https://www.seagate.com/au/en/internal-hard-drives/hdd/barracuda/#specs-3-5

I am sure the other companies are just as easy to get hold of the specs.

[Curufinwe_wins]

28 minutes ago, mr moose said:

Go the products page for specs then.   just like other products with standard model numbers,  the tech changes everytime. 

 

You can check the advertised speeds and spec pretty easy:

https://www.seagate.com/au/en/internal-hard-drives/hdd/barracuda/#specs-3-5

I am sure the other companies are just as easy to get hold of the specs.

Part of the reason why these drives are causing RAID issues is because even in software/firmware, they are not properly reporting their technology. So the array thinks something is totally boned and kicks out a failure.

 

This is a serious issue. On multiple levels.

[mr moose]

2 hours ago, Curufinwe_wins said:

Part of the reason why these drives are causing RAID issues is because even in software/firmware, they are not properly reporting their technology. So the array thinks something is totally boned and kicks out a failure.

 

This is a serious issue. On multiple levels.

There does seem to be an issue with WD red for that, they are specifically nas drives and probably should be noted,  but seagate do make that information available for the NAS Archive drives (I haven't personally looked)  and the reason they don't for they domestic products is because it's of barely noticeable performance issue. As I said, when I buy a new drive, I look at the performance specs,  if the drive is too slow I don't bother SMR or not).   The reality is for many domestic users SMR is probably better because they get more storage for their dollar.

[davide30541]

but is this not a software issue? its an hdd that works, only in a way that nas OS are not expecting. they treat these new hdds as broken because they behave differently from the ones in there already. once nas OS is aware that these drives are diferent this will become a non issue? just thinking here. 

[StDragon]

3 hours ago, Curufinwe_wins said:

Part of the reason why these drives are causing RAID issues is because even in software/firmware, they are not properly reporting their technology. So the array thinks something is totally boned and kicks out a failure.

I've stated this on another thread, in my experience and from others posting online, the array doesn't kick the drive out. Rather, the drive drops on its own without notification. As such, the the array is reporting a failure but not of its choosing.

[wanderingfool2]

13 hours ago, Zodiark1593 said:

Barracuda drives are more likely to be used as boot drives. SMR would be entirely unacceptable in this application as well. 

Comparing a SMR for a consumer drive and NAS one are completely different levels of acceptable though.  The usecase on a NAS level drive, like the red, is unacceptable because the expected workload (in cases of rebuilds and such) are greatly affected by SMR.  I would suspect that the cache on the drives would actually negate a lot of the side-effects of using it as a boot drive (as opposed to one that doesn't).  The reason being you take the penalty when writing data in cases where it needs to be shingled.  Also, I would argue that boot drives require more read operations than writes (which SMR doesn't affect)

 

This would be speculation as I couldn't find any real-world testing of this, but I would suspect they have algorithm's that prevent the overlapping until it is needed.

 

7 hours ago, Unixsystem said:

I don't think it matters if they are Barracudas or if Seagate comes out with a new line called "Garbage Drives For Garbage People," if the drives used the be universally one standard of performance and then it gets changed to a grabbag of maybe worse performance, there needs to be either a rebrand or a concerted effort to inform consumers both on their spec sheet as well as on the various store listing what drive is running what technology. 

Well those models have been out since 2016...so it could be very likely that it was using SMR all this time.  Also, it is different when it is consumer everyday use drive, vs a NAS grade drive.

[Kisai]

12 hours ago, hitardo said:

Guys, I understand your point of view.

But this case is more than that.

 

We are talking of and OEM changing the core of the product, without communicating the change.

 

Same as, a new phone:

Samsung Quiet S, with a Snapdrogn SoC.

Then, after selling millions of those, Samsung changes to:

Samsung Quiet S, with a MediaTek SoC.

 

What would you think of that?

Maybe call it Samsung Quiet S1, or S less.

But never call it the same.

 

Especially when the performance has decreased.

 

The problem is that if they never advertise the technology used, they never have to walk it back. This is what government regulations are for. To avoid these bait-and-switch products from existing.

 

The average consumer only cares about the capacity, not the specs, and the specs don't really mean anything. 

[mr moose]

5 minutes ago, wanderingfool2 said:

Well those models have been out since 2016...so it could be very likely that it was using SMR all this time.  Also, it is different when it is consumer everyday use drive, vs a NAS grade drive.

It's very possible as I have bought them given that I have several of those models purchased over the last few years.  To be honest,  if I do I haven't noticed the speed difference. I use them as storage drives on my main system and to install my games on.

[leadeater]

7 hours ago, Curufinwe_wins said:

Part of the reason why these drives are causing RAID issues is because even in software/firmware, they are not properly reporting their technology. So the array thinks something is totally boned and kicks out a failure.

 

This is a serious issue. On multiple levels.

Would also like to point out that SMR drives can and are used in large storage arrays and RAID, SMR doesn't make a drive not a NAS or RAID drive, not that specific thing anyway.

 

As you said it's a firmware problem and storage controllers or software not being told the drive type so not being able to treat the drive how it would if it knew what it was.

 

An SMR disk can be a NAS/RAID certified drive and work fine like that, just know your workload and make sure the disk model you are buying isn't known to have problems,

[Kilrah]

8 hours ago, mr moose said:

I'm not sure I see a problem here,  Every time I buy a new hard drive (about 1 year on average) I don't assume the one I am getting is identical to one I bought a year or 3 ago.  Technologies and manufacturing processes change.  if you buy an i7 you wouldn't assume it was the same as last gens i7 would you?

But would you be happy to assume it could have half the performance? Normally they get better, not worse.

 

8 hours ago, mr moose said:

Go the products page for specs then. 

The whole issue here is that the change and according performance are NOT documented in the specs.

 

[mr moose]

12 minutes ago, Kilrah said:

But would you be happy to assume it could have half the performance? Normally they get better, not worse.

 

It appears I am very likely using one now that I didn't even realize.     Even if I wasn't I haven't seen the performance numbers, are they actually different to the specs seagate listed for that drive?

 

12 minutes ago, Kilrah said:

The whole issue here is that the change and according performance are NOT documented in the specs.

 

the model numbers change,  they are not the same product.  Just because they look identical and use a rolling model number doesn't mean you should assume a drive released today is exactly the same as one released in 2016.