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

[leadeater]

31 minutes ago, Spotty said:

Who are you meant to buy HDDs from if not Seagate, WD, and Toshiba?

HGST (yea I consider those products different to WD), plus you could also go up a product range from all 3 of those companies which are all PMR only but you'd end up paying more than the product ranges in question pre DM-SMR.

[Spotty]

Just now, leadeater said:

HGST

Western Digital (who owned HGST) phased the HGST brand out a few years ago. HGST no longer exists.

 

7 minutes ago, leadeater said:

plus you could also go up a product range from all 3 of those companies which are all PMR only

But I don't know which ones are SMR and PMR! That's the problem the boycot is supposedly over.

 

The problem isn't SMR drives existing. I think they have a place as low cost bulk storage drives. I don't have any problem with WD/Seagate/Toshiba selling SMR drives. I just want them to list if it's PMR or SMR in the data sheet.

[maartendc]

Wow, this seems unacceptable. Especially if a drive is sold under the same name and may or may not be affected.

 

If they are selling drives that operate differently, they should have a different name. Simple as that. Name them WD Red S or WD Yellow-polkadot or whatever.

 

Manufacturers always try to get away with stuff like this. Just like Nvidia with their lower end GPU's selling less powerful GPU's under the same name.

[Lady Fitzgerald]

4 hours ago, Spotty said:

...Some people are going to say "Buy all SSDs!" but HDDs still have their place thanks to their high capacity/cost...

Even though I am extreme example of those who have gone all SSD, I agree that solution isn't practical for everyone.

[Spotty]

2 hours ago, maartendc said:

Name them WD Red S or WD Yellow-polkadot or whatever.

That's for the 2.5" drives.
It was an itsie bitsie teenie weenie yellow polka dot hard drive

[leadeater]

4 hours ago, valdyrgramr said:

Kinda sad that they phased out HGST.  When I went to get a new platter drive back in like 2017-2018 I went to check that site that does the drive reliability testing, and at the time HGST and Toshiba had the lowest failure rates with WD and Seagate having the highest.   Was considering comparing the prices between an 8TB Toshiba and an HGST, but now I'll just get one of the two Toshiba's I'm looking at.   The N300, or pay 12 bucks more, and get an enterprise drive which has 256mbs of cache.

The HGST products still exist, they are just branded as WD now. But the design of them etc is still all HGST currently.

 

5 hours ago, Spotty said:

The problem isn't SMR drives existing. I think they have a place as low cost bulk storage drives. I don't have any problem with WD/Seagate/Toshiba selling SMR drives. I just want them to list if it's PMR or SMR in the data sheet.

True but as of now the high range products are all PMR only but realistically they cost too much for lower end/home user NAS. Even the smallest step up, to WD Red Pro, is 20% higher cost. Seagate IronWolf and IronWolf Pro are all a higher price than that and it only gets worse from there as you transition out of "Prosumer" and in to actual enterprise products.

[Dylanc1500]

1 hour ago, Spotty said:

That's for the 2.5" drives.
It was an itsie bitsie teenie weenie yellow polka dot hard drive

that she used for the first time today...

 

Thank you, as I will now have that song piquing my mind for the rest of the day.

[mr moose]

11 hours ago, RejZoR said:

And the whole problem literally happened because of this. Do you actually care if drive has EAXF at the end of model number or that it's EEXF ? No, of course not, for RAID you just look if it's of same series (Red) and specs like RPM. This is why this whole controversy and problems even happened in the first place. People looked at the shops, picked WD Red drive based on those specs coz they already had those, stuck them in their NAS or whatever and the thing shit its pants. It's only after all this happened that info came out that "hey, after certain model number, those are actually SMR drives that don't work too well with existing CMR drives". From a 3rd party not vendors themselves! And people who bought a whole new thing from a new batch where they were all SMR's there probably wasn't even a problem where in mixed setups it was. I mean, how often does anyone buys a single WD Red drive for purposes of using it as a single drive? They should've seen this coming and they could avoid it all by just being open about it. People would forget about the fact they are SMR from now on much faster than about not just 1 but ALL the HDD vendors lying about it and hoping they wouldn't get caught. For speeds users would often blame their setups or local problems so I bet they were hoping no one would notice it. Why else they'd be so quiet about it and hiding it everywhere. Hell, you can't even see that from speed specs because they always only state read specs iirc and SMR mostly affects write ones so that also gets hidden within available specs.

The problem occurred because WD released a drive that for whatever reason has not been able to do what they said it would.   It doesn't matter what the technology is inside a product, so long s the product does what it claims on the box.  No one asks or cares what type of fuel injection algorithm a car has or what the main gear of a sewing machine is made from, what they want to know is will it run as expected.   The problem here is not whether the drive has SMR as a listed spec or not, but that the drive is not doing what it is advertised to do.

 

 

[Demonking]

12 hours ago, hitardo said:

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

 

@Demonking apparently, the HDD you mentioned is affected.

 

Model: ST2000DM008 

By the link you posted: https://www.newegg.com/seagate-barracuda-st2000dm008-2tb/p/N82E16822184773?item=N82E16822184773

So that means I can't store and play games on it ?

[RejZoR]

7 hours ago, mr moose said:

The problem occurred because WD released a drive that for whatever reason has not been able to do what they said it would.   It doesn't matter what the technology is inside a product, so long s the product does what it claims on the box.  No one asks or cares what type of fuel injection algorithm a car has or what the main gear of a sewing machine is made from, what they want to know is will it run as expected.   The problem here is not whether the drive has SMR as a listed spec or not, but that the drive is not doing what it is advertised to do.

 

 

I'm pretty sure part of the problem was SMR exactly. When command came in from RAID array to do something, all the other drives were "alright, lets rock on" and SMR drive was like "holup guys, I got to shuffle this shit around"  and then whole thing just shit itself because it's expected that same drives would behave the same at the same time...

[Alen TFG]

OK. After reading all this comments, I will go and see if I can get myself a 2 TB SSD or M.2 hard drive for my pc.

[RejZoR]

4 minutes ago, Alen TFG said:

OK. After reading all this comments, I will go and see if I can get myself a 2 TB SSD or M.2 hard drive for my pc.

If you're using a single drive it doesn't even matter. Problem was with RAID arrays because the drive behaved differently than the rest and that caused a problem.

[leadeater]

19 minutes ago, RejZoR said:

I'm pretty sure part of the problem was SMR exactly. When command came in from RAID array to do something, all the other drives were "alright, lets rock on" and SMR drive was like "holup guys, I got to shuffle this shit around"  and then whole thing just shit itself because it's expected that same drives would behave the same at the same time...

Kind of but it shouldn't cause a rebuild to actually fail. The RAID system, either software or hardware, controls the I/O and issues the commands so if a disk slows down the other disks won't actually care or need to do anything. Rebuilding will happen at the rate of the slowest piece in the chain, slow doesn't cause failure.

 

The disks start off in a high performance state and as the PMR zone fills up you then have to start relying on the firmware of the disk to actually work properly and handle that situation. Once the PMR zone is full it should be going in to passthrough or direct write SMR mode which is slow as shit, usually like 30MB/s to 40 MB/s, but stay operational and responding to I/O commands. These WD Reds are not staying operational, they go in to some weird ass pending state and just stop reporting anything at all while it's trying to cleanup the PMR zone, that's not normal. This is why RAID systems have a shit fit because the disk stops responding, that's a firmware issue.

[mr moose]

1 minute ago, RejZoR said:

I'm pretty sure part of the problem was SMR exactly. When command came in from RAID array to do something, all the other drives were "alright, lets rock on" and SMR drive was like "holup guys, I got to shuffle this shit around"  and then whole thing just shit itself because it's expected that same drives would behave the same at the same time...

 

Maybe it was, but if the drive was labelled as a NAS drive then SMR or not it should have worked.   The issue is presenting a product that either couldn't do as advertised or could only do it under very specific conditions (like all drives being exactly the same in that scenario).  Simply labeling the drive as being an SMR drive would not in and of itself have prevented people from buying them and putting them into raid setups because they were labeled as NAS drives.   The only people who should have thought twice about it would maybe have been storage professionals who might be hesitant to put a slower drive into a performance based setup (remember we aren't talking data centre here, most of these drives are destined for prosumer and small scale situations).   We average users have the benefit of hindsight in this, it is only because someone found out the hard way  that there are problems that we even know about it.  

 

Possible ways to look at it:

 

WD advertised a NAS drive that doesn't work in all raid settings -> Wd's fault

WD put SMR onto drives knowing it wasn't a good tech for NAS -> WD's fault

WD didn't label NAS drives as SMR not believing it would cause an issue -> sad mistake but WD's responsibility

WD Labels drives as SMR, drives fail in raid for unknown reasons -?WD still has to take responsibility because the drive was advertised as a NAS drive.

WD doesn't label and domestic drive as SMR - > no one notices speed difference because that shit varies too much in real world situations and people are buying cheap mass storage, not fast cheap storage or fast mass storage. 

 

As far as I am concerned, there is a case for not labeling drives as being SMR, but there is no case for drives not doing as intended.

 

 

[GDRRiley]

51 minutes ago, Alen TFG said:

OK. After reading all this comments, I will go and see if I can get myself a 2 TB SSD or M.2 hard drive for my pc.

you'll do just fine on some lines of desktop HDD. toshibas X300 is like a slightly cheaper WD black.

[Lady Fitzgerald]

1 hour ago, Alen TFG said:

OK. After reading all this comments, I will go and see if I can get myself a 2 TB SSD or M.2 hard drive for my pc.

M.2 hard drive? There is no such thing as an M.2 hard drive. SSDs can be M.2, U.2, or 2.5" form factors. M.2 is an SSD form factor (physical configuration) that can be either SATA or NVMe.

 

A hard drive (HDD or Hard Disk Drive) is a mechanical drive with moving parts inside. It isn't an SSD (Solid State Drive) which has no moving parts.

[Dabombinable]

My ST2000DM001 uses SMR, and it is just fine in RAID0 with with an ST2000DM006. What I'd be more worried about is having to go into a datasheet to find the cache amount and RPM of a HDD, instead of seeing it clearly on the specifications page. That drive btw came from an external drive (it was hitting over 65^oC - enclosure was quite terrible) which has just shy of 1000 days powered on.

 

Next drive I'm going to be getting though is one targeted at NAS usage, due to the number of HDD in my system (though the other 3.5" is only a 5400RPM F3 - low vibrations from it).

[Dylanc1500]

1 hour ago, Lady Fitzgerald said:

M.2 hard drive? There is no such thing as an M.2 hard drive. SSDs can be M.2, U.2, or 2.5" form factors. M.2 is an SSD form factor (physical configuration) that can be either SATA or NVMe.

Don't forget 3.5", PCIe AIC, Mini PCIe, and my personal favorite EDSFF (better known as the "Ruler" SSD or E1.L). Also, M.2 is a connector form factor that allows NVMe as well as AHCI over the PCIe bus and is commonly used for things such as  WiFi, Bluetooth, satellite navigation, NFC, WWAN and many others. It is also used as a USB and SATA bus connector.

 

Note: You probably honestly already know this, as you are extremely active on the forum, so it's more for the less knowledgeable people that are perusing about and come across this.

[Maticks]

Unraid seems to have less of an issue with SMR than a standard Raid, Provided the Parity drives are CMR.

At least with Unraid and the Data drives only needing to be accessed for the data on them the drive has idle time to rewrite the tracks.

I did have 4 SMR drives in my Synology NAS and they always had latency issues, others reported the drives dropping out.

 

 

[TechyBen]

12 hours ago, mr moose said:

The problem occurred because WD released a drive that for whatever reason has not been able to do what they said it would.   It doesn't matter what the technology is inside a product, so long s the product does what it claims on the box.  No one asks or cares what type of fuel injection algorithm a car has or what the main gear of a sewing machine is made from, what they want to know is will it run as expected.   The problem here is not whether the drive has SMR as a listed spec or not, but that the drive is not doing what it is advertised to do.

 

 

They do. Modders/car nuts do. And these things are listed. Sorry.

 

 

(Aus too, cos you guys are so great at you car stuff )

[TechyBen]

5 hours ago, RejZoR said:

I'm pretty sure part of the problem was SMR exactly. When command came in from RAID array to do something, all the other drives were "alright, lets rock on" and SMR drive was like "holup guys, I got to shuffle this shit around"  and then whole thing just shit itself because it's expected that same drives would behave the same at the same time...

Nonononono. You don't get it, even if the HDD is made of jello, it's suppose to work exactly right for the NAS. It's a manufacturers fault in the firmwares nothing to do with jello being the wrong choice for a NAS drive.

 

12 hours ago, mr moose said:

The problem occurred because WD released a drive that for whatever reason has not been able to do what they said it would.   It doesn't matter what the technology is inside a product, so long s the product does what it claims on the box.  No one asks or cares what type of fuel injection algorithm a car has or what the main gear of a sewing machine is made from, what they want to know is will it run as expected.   The problem here is not whether the drive has SMR as a listed spec or not, but that the drive is not doing what it is advertised to do.

 

 

If Samsung brings out a new HDD that has no RAM cache, but uses a second mini super fast platter and magic super fast 18 read/write arms as a cache, it will perform exactly as a HDD with a RAM cache. However, if *only* Samsungs magic new HDD with super fast cache platters fail, it is *both* the new tech (magic super fast cache platters) *and* the fact it "fails as advertised".

 

You're stating "the average is different" as we are stating "the specific data point is different!" Both are correct, both averages and specific data points are correct. Both the SMR in this drive [slows to a point the OS/NAS/RAID in some systems] fails and the drive fails [some expectation*] of spec.

 

 

*Which might be fixable in software/OS/firmware as with the AMD CCX problems were, but if customers/OS/firmware writers are not told for the NAS servers, then how can they fix it?

[mr moose]

15 minutes ago, TechyBen said:

They do. Modders/car nuts do. And these things are listed. Sorry.

 

 

(Aus too, cos you guys are so great at you car stuff )

Th base maps they are talking about are for their ECU and comes in their package, they have done the work to get or create base maps for their program. that is not the manufacturer's,  But I digress and that is beside the point, there is a but tone of information that is not on the spec sheet that you just don't need to know as a consumer.  And we have yet to determine if SMR is that yet in any of these cases.

 

5 minutes ago, TechyBen said:

Nonononono. You don't get it, even if the HDD is made of jello, it's suppose to work exactly right for the NAS. It's a manufacturers fault in the firmwares nothing to do with jello being the wrong choice for a NAS drive.

 

If Samsung brings out a new HDD that has no RAM cache, but uses a second mini super fast platter and magic super fast 18 read/write arms as a cache, it will perform exactly as a HDD with a RAM cache. However, if *only* Samsungs magic new HDD with super fast cache platters fail, it is *both* the new tech (magic super fast cache platters) *and* the fact it "fails as advertised".

 

You're stating "the average is different" as we are stating "the specific data point is different!" Both are correct, both averages and specific data points are correct. Both the SMR in this drive [slows to a point the OS/NAS/RAID in some systems] fails and the drive fails [some expectation*] of spec.

 

 

*Which might be fixable in software/OS/firmware as with the AMD CCX problems were, but if customers/OS/firmware writers are not told for the NAS servers, then how can they fix it?

Not too sure exactly what you are trying to say here.  but this whole situation is pretty simple:

 

some hard drives failed in a raid and we don't know why.  It might be SMR it might not.

 

Some hard drives in the consumer market are SMR and not labeled.   We don;t know what the performance gap is if there even is one ion real world terms.

 

You claim it is important to know if the drive is SMR or not,  so show me the benchmarks that show SMR drives are distinctly different in a way that can only be represented by labeling them as SMR drives.  If you can;t show me how they differ then you have no grounds to make claims about an SMR label being relevant.

 

The whole problem here is you think there is a problem because SMR was not listed on drives,  when no one knows what the actual problem is let alone if SMR is even a problem then how do you know it is a problem?

[TechyBen]

1 minute ago, mr moose said:

Th base maps they are talking about are for their ECU and comes in their package, they have done the work to get or create base maps for their program. that is not the manufacturer's,  But I digress and that is beside the point, there is a but tone of information that is not on the spec sheet that you just don't need to know as a consumer.  And we have yet to determine if SMR is that yet in any of these cases.

No. Just no. I had a manufacturers ECU specced and remapped. This is not besides the point just because you are wrong. This is incorrect, and you are deciding it is not?

Who decides what I need to know?

Who needs to show SMR is at fault? Can I not decide if I buy a product or not based on what it is?

 

Quote

 

Not too sure exactly what you are trying to say here.  but this whole situation is pretty simple:

 

some hard drives failed in a raid and we don't know why.  It might be SMR it might not.

 

Some hard drives in the consumer market are SMR and not labeled.   We don;t know what the performance gap is if there even is one ion real world terms.

 

You claim it is important to know if the drive is SMR or not,  so show me the benchmarks that show SMR drives are distinctly different in a way that can only be represented by labeling them as SMR drives.  If you can;t show me how they differ then you have no grounds to make claims about an SMR label being relevant.

 

The whole problem here is you think there is a problem because SMR was not listed on drives,  when no one knows what the actual problem is let alone if SMR is even a problem then how do you know it is a problem?

We do. People have tested the SMR. There is a performance gap. It has been tested.

I did not claim it's "important", I claimed consumers asked to know. They wish to know. It *may* be important. The colour is *not* important to me, buy *may* be to others.

 

So to you, SMR is not important, then you decide it must not be to them. The colour of the HDD is not important to you, so you decide the colour is not to others.

 

You don't get to make that decision, and thus RGB SSDs exist. And people asking for PMR exists! (At least while PMR exists in their rigs and there is a choice, when all drives go to SMR we can revisit this discussion, as with the old Tube displays, some benefits some drawbacks... ).

[NineEyeRon]

I think my 4TB Seagate drive is the one on the list. 

 

I have not noticed it being slower vs other drives I have had as they were all pretty old.

 

I also have it in a read heavy/write light application. All the writes are large and very infrequent. So therefore slowness is expected and planned for.

 

I suspect this is probably how it was expected, it without impact on experience with a gain to the user on par with the loss (apple phone slowdown anyone?).

 

There are some of course who are in a different boat, they purchased the drive for a different application compared to those who boot and run mostly from SSDs. Those people should rightly feel aggravated by this news, even if most of us are just angry for the subterfuge rather than any real world harm.

[mr moose]

46 minutes ago, TechyBen said:

 

Who decides what I need to know?

You can decide that, but you can't pretend there is a legitimate reason for needing that information other than you want it.

 

46 minutes ago, TechyBen said:

Who needs to show SMR is at fault? Can I not decide if I buy a product or not based on what it is?

 

We do. People have tested the SMR. There is a performance gap. It has been tested.

Link or citation please. 

 

46 minutes ago, TechyBen said:

I did not claim it's "important", I claimed consumers asked to know. They wish to know. It *may* be important. The colour is *not* important to me, buy *may* be to others.

If it's not important then why do you need to know so badly?

 

46 minutes ago, TechyBen said:

So to you, SMR is not important, then you decide it must not be to them. The colour of the HDD is not important to you, so you decide the colour is not to others.

SMR is only a problem if there is evidence to show it is.  You may as well try to claim the oxide coatings they use is a problem because they haven't told you what they use. Some oxides are better than others so they should have to list it right?   Because it must be important to someone.  No, they don't have to list it and unless you have a reason to demand they list it then trying to argue that it is important to someone isn't a good enough reason.

 

 

46 minutes ago, TechyBen said:

You don't get to make that decision, and thus RGB SSDs exist. And people asking for PMR exists! (At least while PMR exists in their rigs and there is a choice, when all drives go to SMR we can revisit this discussion, as with the old Tube displays, some benefits some drawbacks... ).

I am not making that decision.  I am pointing out that claiming SMR is a spec that is necessary to know without being able to provide evidence to show why (again, that is evidence that there is an issue that can only be overcome/mitigated by labeling the drive as being an SMR drive) does not make it an important or necessary spec to list. 

 

 

Replace SMR in any one of your arguments with any of these components and you will see how pointless the demand is without evidence:

 

-oxidde coating

-size of coil

-magnet core material

-IC manufacturer

-specific code in the bios

-static mitigation architecture

-PS ripple rejection topology in the signal amplifier.

 

 

There are lots of specs we don't know and will never know.  Demanding that they tell me what coating they use because some oxides are not as good and I don't want a sub performing HDD is silly when you can already just not buy a sub performing HDD.  Knowing the oxide layer is causing it to perform slower doesn't suddenly give you any more usable information than you had before.  You already know not to use a ST2000DM001 in a raid and expect top performance or longevity, knowing it has SMR doesn't change that knowledge, finding out it uses an aluminum oxide instead of some other oxide doesn't suddenly make it any less suitable.

 

Here's a bunch of benchmarks of various drives:

https://www.harddrivebenchmark.net/large_drives.html

If SMR genuinely is big enough to cause a serious issue then those drives will fall down the list naturally.