Western Digital's Red 2 - 6TB NAS drives apparently aren't good for NAS use?

[Mihle]

The main problem with the WD Reds is not that they are SMR, it's that their firmware do not report that they are SMR to what they are connected to.

 

This makes the OS on whatever look at the drive as not performing as it should, (I think they expect drives to be CMR unless told by the firmware that it's not) and you get errors of bad drive if you for example rebuild a RAID with one of them. 

[bigmug]

3 minutes ago, leadeater said:

If you have a failed disk and you buy a replacement and that replacement cannot be added to the array because the rebuild always fails I think that is a bit more than a storm in a teacup. You have a NAS, you brought a NAS disk, the disk won't work in your NAS, seems to tick all the boxes to be able to complain about it.

That won't happen if you use the same model of drive. Which you should.
 

[leadeater]

11 minutes ago, jagdtigger said:

No? AFAIK in each drive on one platter one side is "sacrificed" for data needed for the drive to function. Including physical positioning of tracks. This is written to these as part of the manufacturing process which again most be part of the spec.

Performing a low level format created the tracks and sectors, the physical locality of these is defined by the magnetic technology being used i.e. PMR or SMR. You are describing a technology that is being used in the device which yes is valuable information does not dictate that it must be on the product spec sheet. As I said it certainly would be nice to know however do you need to know?

 

Because you need to consider the original point posed. If there was not an issue with these disks failing to rebuild in arrays would you be wanting to know about it now? A problem has been found with a revision of disks which has made you aware of the product and it's changes but what has not been done is a proper root cause of the issue. Until that has actually been done concluding it is because SMR is being used is not proper fault finding.

 

A disk with bad/slow performance would not cause it to be kicked out of an array. As long as the rebuild is progressing and reporting as such it could take 100 years and the disk will not get kicked out and rebuild fail. So either there is something else doing on or people's configurations have a maximum rebuild time set and that is being exceeded.

 

Even before SMR existed disks could have problems rebuilding and the solution back then was to lower rebuild priority and accept the rebuild is going to take 4 days instead of 2 etc. Disks no matter how slow should not be falling out of arrays without an actual reason.

Edited April 20, 2020 by leadeater

[leadeater]

7 minutes ago, bigmug said:

That won't happen if you use the same model of drive. Which you should.
 

That isn't possible in NAS situations. Service lives of these outlast revisions of disks. People are using the same model, WD Red but FRX and FAX revisions of this disk are different.

[jagdtigger]

7 minutes ago, leadeater said:

You are describing a technology that is being used

*sigh*  Look at any specification in electronics. Every single one of them has the manufacturing process in them. End of argument.

7 minutes ago, leadeater said:

If there was not an issue with these disks failing to rebuild in arrays would you be wanting to know about it now?

Yes because i am buying a NAS drive and not a standard one is because i need that extra robustness, reliability, and speed it offers. By silently exchanging CMR to SMR they compromise all three of my reasons and they have skin on their face to ask the same price for the degraded specs.

[bigmug]

NAS does not equal having disks in vulnerable arrays, susceptible to problems like that. I have stand-alone disks in my NAS, mirroring the data manually.

[jagdtigger]

9 minutes ago, bigmug said:

NAS does not equal having disks in vulnerable arrays, susceptible to problems like that. I have stand-alone disks in my NAS, mirroring the data manually.

https://www.westerndigital.com/products/internal-drives/wd-red-hdd

Quote

Designed for RAID environments

 

[Egg-Roll]

54 minutes ago, Master Disaster said:

https://documents.westerndigital.com/content/dam/doc-library/en_us/assets/public/western-digital/product/internal-drives/wd-red-hdd/data-sheet-western-digital-wd-red-hdd-2879-800002.pdf

 

All the text I made bold is incorrect.

 

Besides which you're missing the main point here, they're selling true NAS ready drives alongside these non NAS ready drives to customers willing to spend more money. If that wasn't the case then sure, they could argue that they didn't mislead anyone (they'd probably still lose but would have a stronger case for sure) but as things stand right now, they're selling 2 different products under the same SKU as the same product and one is significantly inferior to the other.

 

The use case for SMR drives is mass cold storage under a WORM situation, random write use is not what the technology was designed to do and as I said earlier, under some situations it can be bad for SMR drives to be used in a random IO situation.

All bold are correct.

 

NASware is still in all reds, has nothing to do with platter tech.

Small nas is still correct, if you think just because the platter tech changed making it unable to be put in a raid then you don't know much about nas tech.

Once again there's more to a nas drive than just the platter.

Still correct even if they where able to put a worst platter tech in than smr, better heads bearings etc all that makes that statement still true.

When they stated reliability they where not talking about the platter but how the drive won't die on you address a year.

Error recovery again isn't about the platter...

 

Now are they selling them side by side, or have they stopped manufacturing the cmr units? That's the real question. Also it's not illegal to sell 2 different platter tech under the same branding, else Ford would get sued for their ev mustang , wd has done nothing wrong, the models are clearly different, if you want to buy based on bigger cache then that's your choice as a consumer w/o raising flags in what's different vs the old 64mb version.

 

The only factual part of your whole post is the last part of in some situations... I've not used smr nor do I plan to deliberately. (My 6tb might be but it's a blue and had worst issues than stupid little smr).

 

If you feel your rights as a consumer have been violated by this change stop whining about it here and contact your local consumer rights... It won't go anywhere, why? Because they've done no wrong smr cmr are tech that was left unlisted, performance numbers are subjective because while smr and synology may not work well, who says those store bought units can tell the difference.  Not to mention most people who buy nas drives will likely be storing videos and playing them back where smr would work fine with minimal impact on performance and if running raid no one will notice likely.

[leadeater]

20 minutes ago, jagdtigger said:

*sigh*  Look at any specification in electronics. Every single one of them has the manufacturing process in them. End of argument.

So? HDDs basically never have, some of the very first exclusive SMR archive disks did. You're asking for a change, something not done before. And again this is for the most part a nice know not a need to know necessarily.

 

I would also like to know but here is the problem points with what people are saying, which is why I'm saying you don't really need to know because the reasoning raised for needing to know is not correct:

• SMR is the cause of the issue: Not necessarily or not by itself.
• SMR cannot be NAS rated or is not suitable for NAS usage: Yes it can be, anything with the right firmware optimization and vibration dampening is suitable for NAS usage.
• SMR is not reliable: Yes it can be, it was design for long term enterprise archive storage so reliability is a highly weighted factor.

 

If you just want to know if a disk is PMR or SMR or DM-SMR/PMR+SMR then great, more info is always nice. If you want to know because you have a high write workload then even better, you likely do need to know. If you want to know just because you think SMR cannot be used in a NAS you do not need to know and is why manufacturers are hesitant to put that info out because of these bad assumptions.

 

 

20 minutes ago, jagdtigger said:

Yes because i am buying a NAS drive and not a standard one is because i need that extra robustness, reliability, and speed it offers. By silently exchanging CMR to SMR they compromise all three of my reasons and they have skin on their face to ask the same price for the degraded specs.

SMR does not have to change reliability nor does it have to change robustness. Seagate Archive v2 has the same yearly write workload rating as the WD Red (FRX) and similar MRBF, main difference is half the Load/Unload cycles. Many assumptions about SMR, what you think and what is are different things.

 

A PMR+SMR disk can perform the same as a PMR disk if designed correctly.

[jagdtigger]

4 minutes ago, leadeater said:

HDDs basically never have, some of the very first exclusive SMR archive disks did. You're asking for a change, something not done before.

You are just contradicting yourself right there...

 

Same name, same price, hidden change to "inferior" technology and thus degraded performance -> intentional misleading. There is no excuse for this despicable tactic.

Edited April 20, 2020 by jagdtigger

[leadeater]

Just now, jagdtigger said:

You are just contradicting yourself right there...

Ah.. basically i.e. not always. Statement of commonality, highlighting an exception. An exception is not the common which is the point of what was said. Changing so every HDD spec sheet list recording technology used is a change to the commonality.

[jagdtigger]

2 minutes ago, leadeater said:

Ah.. basically i.e. not always. Statement of commonality, highlighting an exception. An exception is not the common which is the point of what was said. Changing so every HDD spec sheet list recording technology used is a change to the commonality.

It wasnt done before because there was only CMR, now that there is SMR its time to update spec sheets because these two technology has some huge differences. Right now what WD does is intentional misleading for financial gains.

[leadeater]

2 minutes ago, jagdtigger said:

It wasnt done before because there was only CMR, now that there is SMR its time to update spec sheets because these two technology has some huge differences. Right now what WD does is intentional misleading for financial gains.

 

15 minutes ago, leadeater said:

If you want to know just because you think SMR cannot be used in a NAS you do not need to know and is why manufacturers are hesitant to put that info out because of these bad assumptions.

 

[TechyBen]

1 hour ago, jagdtigger said:

No? AFAIK in each drive on one platter one side is "sacrificed" for data needed for the drive to function. Including physical positioning of tracks. This is written to these as part of the manufacturing process which again most be part of the spec.

I think the difference between the drives chip/firmware/controller vs the OS. The os (these days, hence the old "track/sector" terminology) no longer knows what shape/position the data is in. With the exception of RAID needs, TRIM needs and some caching such as Optain.

 

I also had a lemon of a Crucial SSD that the controller/cache (IIRC was a dramless cacheless version) died after 15mins use after every format, probably for similar reasons of being specked below min requirements to exist. XD

[TechyBen]

31 minutes ago, leadeater said:

 

 

 

Nope. IMO intentional misinformation. "These new drives *won't* sell because they are carp in a tin sandwiches with mud in them... quick, forget to update the spec sheets! Problem solved".

 

I've been in meetings where vital information was not added... not because they were especially worried customers would make the "wrong" decisions, but that they would make the right ones. (Samsung was the funniest to do this, Hotpoint [Wirlpool] Were a little less funny... just more tragic).

 

Your totally missing the point. Example

Customer "The *red* paint is toxic... argh...

You: "The commonality of the colour was never a problem, you don't know that the colour is the cause of the toxicity, it might be a bad batch or a new formular..."

Customer: Point is it's toxic, and they never told me!!!

 

That's the point. If the cache, controller chip, or SMR is the cause, it does not matter. It's a material change not noted in the spec sheet.

 

I've attempted a repair on a case to a HP laptop, and had to send the parts back, because although all part numbers matches, and were externally identical, they change the positions/design internally. Why? Well, the hinges always broke (hence me needing to repair it!) and half way through an existing model run, they updated the design, but none of the model numbers (externally anyhow, a couple of internal stickers were different). Thing is, the *spec* was identical, and use case, just the physical reliability increased.

 

Here, the actual performance and drive behaviour is different. And not listed!

[leadeater]

5 minutes ago, TechyBen said:

Nope. IMO intentional misinformation. "These new drives *won't* sell because they are carp in a tin sandwiches with mud in them... quick, forget to update the spec sheets! Problem solved".

No you think they are based on a few reports without any explanation as to actually why the problems are happening. Could be spaces ghosts causing the issue, I haven't done the required and sufficient testing to properly diagnose the issue but I'm definitely sure it's space ghosts.

 

DM-SMR will not just drop out of arrays or cause rebuild failures because they are slow. Even then if you don't exceed the write capability of the caching performance is actually better than the old model.

 

Edit:

Also it's still a disk with NAS optimized firmware rated for a set write workload per year in line with a NAS disk and is designed for systems of less than 8 disks. It's not misinformation to market a disk as a NAS disk if it actually is a NAS disk, which is it.

 

Plus some of the reports are coming from people with deployments larger than 8 disks and not using Red Pros, more disks in a disk group puts more stress on the drive recovering which makes it more likely to push a DM-SMR disk over the edge and in to slow mode. Those people don't really have much of a legitimate complaint if they are operating outside the manufacturers recommendations. Honestly in the past with the FRX and older Reds it didn't actually matter, you could happily use 24 in a single system and only suffer a very slightly increase in failure rate from not have the full vibration compensation the Red Pros have.

[Egg-Roll]

2 minutes ago, TechyBen said:

Nope. IMO intentional misinformation. "These new drives *won't* sell because they are carp in a tin sandwiches with mud in them... quick, forget to update the spec sheets! Problem solved".

That would require thinking, something I wonder if Seagate can even do. 

 

Also look at the rn number of the new smr drives, Google that and tell me where in the spec sheets state the platter tech. It's the same for all drives. 

[TechyBen]

1 minute ago, leadeater said:

No you think they are based on a few reports without any explanation as to actually why the problems are happening. Could be spaces ghosts causing the issue, I haven't done the required and sufficient testing to properly diagnose the issue but I'm definitely sure it's space ghosts.

 

DM-SMR will not just drop out of arrays or cause rebuild failures because they are slow. Even then if you don't exceed the write capability of the caching performance is actually better than the old model.

See my update.

 

They made a material change to a device/spec that needs identicallity, and have they listed "V2" or "With SMR"? It's kinda worrying. Especially for a HDD manufacture, who knows batch details are important to customers.

 

As said, a bit like changing paint colour/chemical/type design half way through a batch.

 

LOTS of companies officially "gimp" the current gen/mid tier product with new cheap manufacturing tech without telling customers to save on costs/improve margin, then add the existing old but more reliable tech to the top teir and just up the price/rebrand.  At least Samsung tell customers QVO/EVO/TLC in their SSD so you know if it's new/old/slow/fast/reliable/new etc.

[TechyBen]

5 minutes ago, Egg-Roll said:

That would require thinking, something I wonder if Seagate can even do. 

 

Also look at the rn number of the new smr drives, Google that and tell me where in the spec sheets state the platter tech. It's the same for all drives. 

Oh, I don't disagree there, but this is a first for changing such tech within a model run? Just because "we always did it this way" does not mean you run the customers off a cliff because "in the last 20 year we never had to apply breaks and stop crashing and drowning in the sea".

 

Poor move by these manufactures. Adding a different drive behaviour, without making it a new model number. So at least those buying for NAS could swap out entire arrays etc... that might even make them more cash... but no, they gotta do things on the cheap/can't be bothered route.

 

PS, so you saying AMD was ok in their half compute chips they got sued for?

[leadeater]

5 minutes ago, TechyBen said:

They made a material change to a device/spec that needs identicallity, and have they listed "V2" or "With SMR"? It's kinda worrying. Especially for a HDD manufacture, who knows batch details are important to customers.

Disks do not need to be identical and the disk spec were updated FRX vs FAX. Disks of different revisions with different drive cache sizes are in arrays all the time, disks change over time and you can't always get the same one you had 4 years ago.

 

Edit:

Case in point our Netapp filer used for backups had 3 different brands of 3TB disks in it and at least 10 different model revisions across those brands. All the disks were certified Netapp disks with Netapp firmware replaced by Netapp technicians.

[TechyBen]

1 minute ago, leadeater said:

Disks do not need to be identical and the disk spec were updated FRX vs FAX. Disks of different revisions with different drive cache sizes are in arrays all the time, disks change over time and you can't always get the same one you had 4 years ago.

But you are listing known speccs... You are comparing apples to oranges. These drives are failing. Customers are angry, because it appears the failure is due to a difference, not a fault.

 

We need companies to be more honest and up front. Putting a white paper up will not confuse customers shopping in Walmart.

[leadeater]

4 minutes ago, TechyBen said:

without making it a new model number

[Egg-Roll]

Just now, TechyBen said:

Oh, I don't disagree there, but this is a first for changing such tech within a model run? Just because "we always did it this way" does not mean you run the customers off a cliff because "in the last 20 year we never had to apply breaks and stop crashing and drowning in the sea".

 

Poor move by these manufactures. Adding a different drive behaviour, without making it a new model number. So at least those buying for NAS could swap out entire arrays etc... that might even make them more cash... but no, they gotta do things on the cheap/can't be bothered route.

Technically it's its own model number(fax), but I know what you mean something like red smr.

This out lash towards smr reminds me of Ford using its 4 cylinder eco boost instead of its normal v6 in the f150 (and I think mustang), it'll die down people who need cmr will keep to them, those using nas drives for typical use (video storage, large file storage) will buy smr and likely not see a difference

[leadeater]

5 minutes ago, TechyBen said:

But you are listing known speccs... You are comparing apples to oranges. These drives are failing. Customers are angry, because it appears the failure is due to a difference, not a fault.

And the fault is not proven to be the change to DM-SMR, it's more likely a firmware issue. Firmware issues could happen at any point on any revision on any magnetic storage technology.

 

You're just parroting what you've heard without having the knowledge and proof to know if it is what you say it is and not something else. WD could come out with a firmware fix tomorrow that resolves this and it would look pretty dumb because that would mean DM-SMR was not the issue.

 

If you want honesty then apply that to yourself too, companies are reluctant to give out information you don't need that isn't actually necessary that people may avoid the product or try and hold them to years later when they want to progress the product line. Forcing them to dump the Red product branding simply because you are forcing them to only sell PMR disks under that doesn't actually help you, the Red product line will just end and a new one will start and you still have the same options as you did before.

[TechyBen]

1 minute ago, leadeater said:

OK. This is different. People were arguing over them being the same.

Still a poor move IMO over not noting the tech (see helium or microwave versions in the product line). Or things like number of platters, vibration load, etc etc.

 

Things such as the cache size, being different, for SMR, the seek/write times etc. Do make a difference, and only listing "its a drive!" is a poor move.

 

So thanks, having different model numbers is less misleading. Still a poor move. (Another example is say OLED in TVs/phones, if one manufacture, changes the model number by 1 digit/letter, then drops the OLED for LCD and backlight, and just "forgets" to put anything in the spec sheet listing "it's a display" and not the display tech). Poor move, though not illegal.

 

If someone is a total unreasonable and horrible person, that's not illegal and totally their right. But they may not get invited to my BBQ, may not get presents, and I may stop buying HDDs off them. (Hence why I get Samsung SSDs not crucial, and Toshiba HDDs not Seagate. ).