Western Digital gets sued for sneaking SMR disks into its NAS channel

[leadeater]

26 minutes ago, jagdtigger said:

FreeNAS, and they actually sell neat boxes that fit for consumers just like any other NAS box. Busted......

That's still a singular like I said. I can prove ZFS not being fit for purpose using WD Red FAX because only it has a problem with that hardware. When you are the odd one out you are the problem.

 

Yes this is a stupid argument but I can prove it using this method, it doesn't make it correct, literally the point here.

 

Edit:

I.e. You cannot prove WD Red FAX as not being fit for purpose using ZFS alone.

[jagdtigger]

Just now, leadeater said:

That's still a singular like I said. I can prove ZFS not being fit for purpose using WD Red FAX because only it has a problem with that hardware. When you are the odd one out you are the problem.

And at the same time my above example shows that ZFS is in fact used in NAS systems making these NAS drives unusable in their intended environment. And this where this ends.

[leadeater]

7 minutes ago, jagdtigger said:

And at the same time my above example shows that ZFS is in fact used in NAS systems making these NAS drives unusable in their intended environment. And this where this ends.

Like I said that doesn't prove that the WD Red FAX product is not fit for purpose as a NAS disk, it only proves it is not suitable for use with ZFS. You proved nothing on that. If I go out and buy 100 NAS's from different vendors with 8 or less bays, and I'll be generous here, 10 of them are ZFS based then only 10% would exhibit an issue and the commonality is ZFS. The other 90% that were not ZFS had no issue, so what exactly was actually proven here?

 

Only that WD Red FAX is not suitable for usage with ZFS, not that WD Red FAX is not fit for purpose as a NAS disk. The world is not ZFS.

[mr moose]

45 minutes ago, Master Disaster said:

Depends on the wording of the suit. If they try to sue for false advertising they'll probably lose, if they go for misleading customers they'll probably win.

Misleading customers is false advertising (at least that is specifically spelled out in Australian law and the US law seems to be the same just without those very specific words).  False advertising does not have to be in a specific ad, it can be from any form of communication from the manufacturer.  Under Section 43(a) of the Lanham Act, false advertising is any misleading or false statements, deception or something with the tendency to decieve about a good or service.

 

In this case, I really think the plaintiff is going to have a hard time proving WD mislead them when they made no false claims and their products work as intended in everything except in one software environment.   Just to use a horrid analogy, if shell produced a fuel that worked in every car except Hyundia's,  you wouldn't say it was an issue with the fuel, you'd say it was an issue with Hyundai's.  Given that just like shell can't control how Hyundai tune their cars, WD can't control how freenas tune ZFS. 

 

[leadeater]

1 hour ago, mr moose said:

In this case, I really think the plaintiff is going to have a hard time proving WD mislead them when they made no false claims and their products work as intended in everything except in one software environment. 

But you can angle it that having been given the information they would have not purchased the product and gone with something else, so being mislead in to buying something they otherwise would not have.

 

The stupid thing is the damages they are seeking is only a refund for the product, and obviously legal fees if they win but that's a by product of a suit. The other things they are demanding is WD stop advertising it as NAS ready and also their distributors and retailers, because their single use case of ZFS is obviously representative of the entire NAS market. Literally just some salty people, likely using the product out of spec anyway, stamping their feet demanding something WD was already offering, a refund and free upgrade to WD Red Pro or Ultrastar (or Gold, I forget but all offered options were good), and that their usage represents an entire market.

[Master Disaster]

2 hours ago, mr moose said:

Misleading customers is false advertising (at least that is specifically spelled out in Australian law and the US law seems to be the same just without those very specific words).  False advertising does not have to be in a specific ad, it can be from any form of communication from the manufacturer.  Under Section 43(a) of the Lanham Act, false advertising is any misleading or false statements, deception or something with the tendency to decieve about a good or service.

 

In this case, I really think the plaintiff is going to have a hard time proving WD mislead them when they made no false claims and their products work as intended in everything except in one software environment.   Just to use a horrid analogy, if shell produced a fuel that worked in every car except Hyundia's,  you wouldn't say it was an issue with the fuel, you'd say it was an issue with Hyundai's.  Given that just like shell can't control how Hyundai tune their cars, WD can't control how freenas tune ZFS. 

 

I did some research myself and you are indeed correct.

 

My confusion comes from the fact that most of the world calls it false advertising while here in Europe our courts call it misleading advertising however the basic premise is almost identical between the 2.

 

U.S.C.A Definition - https://legal-dictionary.thefreedictionary.com/false+advertisement

EU Definition - https://legal-dictionary.thefreedictionary.com/misleading+advertisement

 

[wanderingfool2]

9 hours ago, leadeater said:

Well the thing is for the one person so far to actually do proper testing to see how this was affected it actually wasn't by a much. For a 90%-95% full array with mixed drives a rather under resourced lower spec NAS had rebuild times 60% slower and the high spec NAS has rebuild times 40% slower. Now where this will, and why this will fail in court, is that the actual times are not unreasonable and 10TB+ disks that are CMR/PMR have longer rebuild times than 6TB DM-SMR disks (smaller size even quicker) so it's going to fail the customer impact assessment.

 

It would have as much chance at success as suing because WD Red (FRX) has slower rebuild times than WD Red Pro because they sold 'NAS rated disks' with only ~5400 RPM and I the customer deem that to not be 'NAS rated'. Or suing because a 10TB disk takes longer to rebuild than a 6TB disk. Any and all technical arguments are going to be so easily countered and brushed away by a mountain of counter technical evidence and it'll just stall the case until the court just dismisses it.

 

Literally the only case they have here is the non-disclosure of DM-SMR, anything else is irrelevant and will fail. Opinions on DM-SMR and it's NAS suitability just don't matter, for this.

Do you actually have a link to the person who did the testing.  It would interesting to see; with that said introducing a technology into NAS drives at even lets say 20% slower and not advertising it (while keeping the same brand name, and only changing the model number) is deceptive advertising.  40% slower is a huge difference in rebuild times.

 

Comparing 10TB+ disks to lower TB disks rebuild times is a moot point though in this case; since there are quite a bit of people who avoid larger capacity drives due to rebuild times and it is clearly a logical step to determine a 10TB drive will take longer than 6TB.  (Larger drives notably have larger risks carried with them).  Take a 6TB red and compare it to the 6TB red pre SMR and if there is a sizable difference, then yes there would be a customer impact issue.

[leadeater]

4 hours ago, wanderingfool2 said:

Do you actually have a link to the person who did the testing.  It would interesting to see; with that said introducing a technology into NAS drives at even lets say 20% slower and not advertising it (while keeping the same brand name, and only changing the model number) is deceptive advertising.  40% slower is a huge difference in rebuild times.

 

Comparing 10TB+ disks to lower TB disks rebuild times is a moot point though in this case; since there are quite a bit of people who avoid larger capacity drives due to rebuild times and it is clearly a logical step to determine a 10TB drive will take longer than 6TB.  (Larger drives notably have larger risks carried with them).  Take a 6TB red and compare it to the 6TB red pre SMR and if there is a sizable difference, then yes there would be a customer impact issue.

Yes, all the videos are in the original topic about this, it's about 5 video series going through new array setup and rebuild to full data array rebuild with mixed disks and all FAX.

4 hours ago, wanderingfool2 said:

40% slower is a huge difference in rebuild times.

All times were less than 24 hours, 40% looks big if you remove the context.

 

4 hours ago, wanderingfool2 said:

with that said introducing a technology into NAS drives at even lets say 20% slower and not advertising it (while keeping the same brand name, and only changing the model number) is deceptive advertising

The new revision is actually faster for a lot of usage profiles, it only shows significant performance difference during rebuilds.

 

You can still use these disks perfectly fine in your QNAPs and Synology's of the world and be able to rebuild in acceptable times and get the performance benefit of the 256MB disk cache.

 

Like I said there are better options in the market within this price segment and people should be buying those instead but it is flawed to take a single use case/situation as a way to say something is not fit for purpose. With that you can easily prove most products are not fit for purpose, it's not that hard to find situations in a market with an issue or engineer one to get the result you want and then to use that to declare the product is not fit for the entire market is just silly.

 

Rebuild times are not the be all and end all of everything, that is also silly. If it is high enough where it actually is a problem, less than 1 day to 9 days, then it is a problem.

 

Edit:

FYI on engineering a result to prove a bad point, it takes about 3 days to rebuild a 3TB NL-SAS 7200 RPM disk in our Netapp that is used for backups because of the consistent load placed on the disks. 3 days is a very long time for a 3TB disk, it doesn't make it not fit for purpose so I'd never use this to try and argue that sort of line.

 

The additional context that needs to be applied here is that each disk shelf is configured in triple parity with two hot spares and there are two Netapps in different cities and data is replicated to both, risk of data loss due to disk failures is very low and these longer rebuild times are not a significant issue.

[mr moose]

9 hours ago, leadeater said:

But you can angle it that having been given the information they would have not purchased the product and gone with something else, so being mislead in to buying something they otherwise would not have.

Even if they reverse the angle, they are still going to have to prove that SMR created a condition that caused unwarranted outcomes.  Otherwise knowing it was there or not is moot.   The best they can do is show it is a bit slower on full rebuilds in comparison to another drive that does not advertise a speed in which that would take place.  No hdd manufacturer has ever spec'd how long a rebuild should take, only that it will.

 

9 hours ago, leadeater said:

The stupid thing is the damages they are seeking is only a refund for the product, and obviously legal fees if they win but that's a by product of a suit. The other things they are demanding is WD stop advertising it as NAS ready and also their distributors and retailers, because their single use case of ZFS is obviously representative of the entire NAS market. Literally just some salty people, likely using the product out of spec anyway, stamping their feet demanding something WD was already offering, a refund and free upgrade to WD Red Pro or Ultrastar (or Gold, I forget but all offered options were good), and that their usage represents an entire market.

Agree.

[Curufinwe_wins]

ZFS is not the only time these drives are completely unsuited and unsuitable for NAS use, only one of the most dramatic ones.

 

Good luck consistently rebuilding any striped array with an SMR-only system. Not to mention the 10-100x access penalties for more random operations. NAS does not mean sequential only. Nor does it mean read-only. And within WDs OFFICIAL specs, you can easily thrash/trash these drives. 

 

People saying ZFS is the issue are completely missing the point.

[divito]

12 hours ago, jagdtigger said:

And at the same time my above example shows that ZFS is in fact used in NAS systems making these NAS drives unusable in their intended environment. And this where this ends.

Entry-level power tools exist, and using them vs. a higher-spec'd tool might make certain jobs take longer, but it doesn't make those tools unfit.

And as leadeater pointed out, if I bought these specific drives, and then chose ZFS and found a problem, but then I switched to something else, then I guess ZFS is the problem. ZFS isn't fit for NAS application. 

That's just horrendous logic. As with all technology, some things have better compatibility than others. Given the immense variability in hardware and software, there are bound to be conflicts and anomalies, but given that ZFS is very small part of the storage world, it does not prove these drives are unfit.

[Curufinwe_wins]

9 minutes ago, divito said:

Entry-level power tools exist, and using them vs. a higher-spec'd tool might make certain jobs take longer, but it doesn't make those tools unfit.

And as leadeater pointed out, if I bought these specific drives, and then chose ZFS and found a problem, but then I switched to something else, then I guess ZFS is the problem. ZFS isn't fit for NAS application. 

That's just horrendous logic. As with all technology, some things have better compatibility than others. Given the immense variability in hardware and software, there are bound to be conflicts and anomalies, but given that ZFS is very small part of the storage world, it does not prove these drives are unfit.

Clearly consumer NAS has to mean not ZFS or any other consumer striped array system. That makes total sense.

 

/sarcasm

[mr moose]

15 minutes ago, Curufinwe_wins said:

ZFS is not the only time these drives are completely unsuited and unsuitable for NAS use, only one of the most dramatic ones.

 

Good luck consistently rebuilding any striped array with an SMR-only system. Not to mention the 10-100x access penalties for more random operations. NAS does not mean sequential only. Nor does it mean read-only. And within WDs OFFICIAL specs, you can easily thrash/trash these drives. 

 

People saying ZFS is the issue are completely missing the point.

I believe that random operations are quicker with DM SMR drives due to the larger cache.  It's only the build time that suffers.

 

That is the issue here, independent tests have shown the only draw back seems to be rebuild time for a large array or ZFS.

 

There is no evidence they are slower at anything else or perform notably less in day to day usage.   Which leads me to my earlier comments that if they were observably slower in day to day usage then a lot more people should have worked it out when they bought them, not a year later when they were told.

[TetraSky]

They really should've labelled their drives. This is the consequence of that.

[Curufinwe_wins]

16 minutes ago, mr moose said:

I believe that random operations are quicker with DM SMR drives due to the larger cache.  It's only the build time that suffers.

 

That is the issue here, independent tests have shown the only draw back seems to be rebuild time for a large array or ZFS.

 

There is no evidence they are slower at anything else or perform notably less in day to day usage.   Which leads me to my earlier comments that if they were observably slower in day to day usage then a lot more people should have worked it out when they bought them, not a year later when they were told.

Random operations are quicker for the first small bit then absolutely evaporate after that (WD has a white paper on it actually). Firmware updates have helped some, but I can show some hilariously bad results from earlier generation SMR drives with similar caches.

 

Further compounding the issue is that WD explicitly misrepresents in it's firmware the drive technology so the OS, which literally no matter how you cut it, is in fact misleading/false advertisement. 

 

Their sales representatives and service handlers also misrepresented the products on numerous recorded occasions.

 

SMR drives also have signficant issues as the drive approaches full (just as psuedo-SLC hard drives do). 

 

These performance characteristics are blatantly apparently not ok for NAS recommendations, and thats why both WDs staff and every single other HDD maker doesn't recommend them and said they are unsuitable for that use case.

 

Consumers have been reporting issues with it for months, but it took systematic evaluation yo get to the bottom of it, explicitly because of how much denial and misrepresentation WD did up front about it.

 

They have literally no excuse for this.

[mr moose]

41 minutes ago, Curufinwe_wins said:

Random operations are quicker for the first small bit then absolutely evaporate after that (WD has a white paper on it actually). Firmware updates have helped some, but I can show some hilariously bad results from earlier generation SMR drives with similar caches.

 

Further compounding the issue is that WD explicitly misrepresents in it's firmware the drive technology so the OS, which literally no matter how you cut it, is in fact misleading/false advertisement. 

Did you watch the video's posted in the other thread?

EDIT: just as a n after thought to this,  WD's white paper would be on earlier versions of SMR and not their current DM SMR?  It is a little difficult to believe they claim specific limitations where they don't exist in testing.

Quote

Their sales representatives and service handlers also misrepresented the products on numerous recorded occasions.

Which ones?

Quote

SMR drives also have signficant issues as the drive approaches full (just as psuedo-SLC hard drives do). 

 

These performance characteristics are blatantly apparently not ok for NAS recommendations, and thats why both WDs staff and every single other HDD maker doesn't recommend them and said they are unsuitable for that use case.

Again,did you watch the video's where they were tested in the advertised end use?

 

Quote

Consumers have been reporting issues with it for months, but it took systematic evaluation yo get to the bottom of it, explicitly because of how much denial and misrepresentation WD did up front about it.

 

They have literally no excuse for this.

 

Consumers report issues about all products all the time,   There hasn't been anything specific to these drives in raid before or after the ZFS issue that makes this worse than any other product out there. 

[leadeater]

2 hours ago, Curufinwe_wins said:

but I can show some hilariously bad results from earlier generation SMR drives with similar caches

Early generation SMR disks were marketed and sold as archive disks and were also not recommended to be put in RAID or to be used as primary storage. I wouldn't assume nothing has changed between those disks and these ones.

 

For current testing see here (plus the other videos in the series)

 

TL;DR Basically no different in writing reasonably large data sets, writing ~20GB then writing again another ~20GB by extracting the zip files. You'll probably see the performance drop but in a 4 bay NAS's you'll have to write in a single session about 100GB-150GB if not more. It also makes no difference if the amount of data written is random I/O or partly random or sequential, the PMR zone is a fixed size and acts like any other PMR disk. Where you get the problem is when this hits high water mark from the disk not being idle enough or the rate of data in exceeds the rate of data out causing it to get full and hit the high water mark.

 

If you have sustained high write loads on your NAS with little to no idle times then these disks are not something you should use, nobody should be buying them anyway with other better market options.

[leadeater]

1 hour ago, mr moose said:

Did you watch the video's posted in the other thread?

No I don't think anyone has. "DM-SMR is bad mkay", "I haven't actually looked for real data to show this is still the case or the situations where it is actually in fact slower but I don't care, it's just bad all the time always because that's what I believe"

 

The problem isn't that these can be slower, because that is true, nor is it that they can be faster, that is also true. It's the ludicrous claims (not here, the Reddit ones) not backed by evidence that aren't true, because if that evidence existed for these disks you could simply just show it and not have to use hyperbolic statements and use real facts and figures.

 

We have real facts and figures now, one set of testing shows very clearly that these disks should not be used with ZFS under any situation and the other set of testing shows minimal difference between the previous revision and the current revision each with their own situational performance leads.

[Curufinwe_wins]

1 hour ago, leadeater said:

Early generation SMR disks were marketed and sold as archive disks and were also not recommended to be put in RAID or to be used as primary storage. I wouldn't assume nothing has changed between those disks and these ones.

 

For current testing see here (plus the other videos in the series)

 

TL;DR Basically no different in writing reasonably large data sets, writing ~20GB then writing again another ~20GB by extracting the zip files. You'll probably see the performance drop but in a 4 bay NAS's you'll have to write in a single session about 100GB-150GB if not more. It also makes no difference if the amount of data written is random I/O or partly random or sequential, the PMR zone is a fixed size and acts like any other PMR disk. Where you get the problem is when this hits high water mark from the disk not being idle enough or the rate of data in exceeds the rate of data out causing it to get full and hit the high water mark.

 

If you have sustained high write loads on your NAS with little to no idle times then these disks are not something you should use, nobody should be buying them anyway with other better market options.

I admitted that obviously firmware improvements have helped...

 

But again... any time any striped system rebuilds (or large backups in general)... as was one of my initial comments.  And the assumption that someone should have to go away from nas drives to build a nas is pretty insane IMO. 

 

The previous WD Red drives (well the CMR ones) were perfectly good enough for that purpose. Personally, my drives are from just before WD killed the separate HGST line, but still at least the He drives give a great price to perf overall. 

 

This gets back to something we talked about in a previous thread IMO. If the specs weren't intentionally deceptive. If WD hadn't been intentionally deceptive. If the labels weren't intentionally obtuse... I wouldn't care. If they made a lineup called Red SMR or something that explicitly told non-expert people anywhere that there was this risk in using these drives that are marketed for this explicit purpose, I wouldn't be so pissed about this. But they didn't, and deserve every single one of these lawsuits shoved down their throat.

[leadeater]

1 minute ago, Curufinwe_wins said:

But again... any time any striped system rebuilds (or large backups in general)... as was one of my initial comments.  And the assumption that someone should have to go away from nas drives to build a nas is pretty insane IMO. 

And in the series of videos linked that was tested and the rebuild times were well within reason, unlike ZFS rebuilds which are well outside of reason.

[Curufinwe_wins]

2 hours ago, mr moose said:

Did you watch the video's posted in the other thread?

EDIT: just as a n after thought to this,  WD's white paper would be on earlier versions of SMR and not their current DM SMR?  It is a little difficult to believe they claim specific limitations where they don't exist in testing.

Which ones?

Again,did you watch the video's where they were tested in the advertised end use?

 

 

Consumers report issues about all products all the time,   There hasn't been anything specific to these drives in raid before or after the ZFS issue that makes this worse than any other product out there. 

Ok so let me be blunt. We hashed this out in great detail in the original thread about the issue being brought to awareness, and I honestly don't have the time nor inclination to dredge back up all of the sources and evidence discussed at that time. If that means you disbelieve the previous discussion, I understand, even if it's annoying.

 

I do have this bookmarked ofc, and it's from around that time. It shows at least a few random incorrect employee statements and other situations. 

 

https://arstechnica.com/gadgets/2020/04/caveat-emptor-smr-disks-are-being-submarined-into-unexpected-channels/

[Delicieuxz]

Edited for clarity:

 

I wonder what Linus Torvalds would say about the argument of 'if a design change breaks the functionality of an odd-one-out, then the problem is the odd-one-out'.

 

https://lkml.org/lkml/2012/12/23/75

 

Specifically: "WE DO NOT BREAK USERSPACE!"

 

And: "The fact that you then try to make *excuses* for breaking user space, and blaming some external program that *used* to work, is just shameful. It's not how we work."

[Shreyas1]

What I wonder is, will we need specialized judges for these types of cases in the future?

 

I mean, my impression is that the avg judge probably doesn't even know how exactly a hdd works or what NAS stands for, much less about different hdd technologies and such.

 

In the future, as these cases will become more popular, will we need specialized "tech judges" who know more about these things and will be able to make a fairer decision? Or will this negatively impact our justice system in any way?

 

just a thought

[Curufinwe_wins]

12 minutes ago, leadeater said:

And in the series of videos linked that was tested and the rebuild times were well within reason, unlike ZFS rebuilds which are well outside of reason.

Hard to say that a 50% reduction in performance for moderate file copies is reasonable for what is marketed by spec as a pure performance improvement.

 

(Note I HATE watching videos when articles work just as well, I hadn't watched the video, but had read the STH articles about it. And a number of others in the past ofc closer to when the issue was first getting attention.)

 

https://www.servethehome.com/wd-red-smr-vs-cmr-tested-avoid-red-smr/2/

 

This isn't some insane transfer. It's a NAS. It preforms backups and bulk storage. That is not ok for a drive that if you only look at WDs marketing and specs looks purely better than the CMR equivalent model.

[leadeater]

1 minute ago, Delicieuxz said:

I wonder what Linus Torvalds would say.

 

https://lkml.org/lkml/2012/12/23/75

Ah yes because everything he says is gospel and the way he conducts himself like that should be mirrored by everyone, also that post is 100% relevant here 

 

It's not like these new revisions wouldn't have a purpose if they were actually cheaper, then I might have some product metric to use to potentially recommend them but they aren't cheaper so they make no sense at all. Like in the other topic if it was deployed in a 10TB+ disks and they were 30% cheaper then great they would actually have a reason to exist.