Western Digital binds up self-inflicted SMR wound

Yorick

Wizard
Joined
Nov 4, 2018
Messages
1,912
If you’ve been living under a rock: WD Red 2TB through 6TB became SMR drives in their EFAX variant, which is unsuitable for use with ZFS.

After the considerable blowback this created, WD have bound up this wound by bringing back the EFRX (CMR) 2-6TB drives and creating a clear branding distinction.

WD Red: All SMR drives
WD Red Plus: All CMR drives (this is the original all-CMR Red lineup, before SMR was introduced into the Red line)
WD Red Pro: Unchanged, all CMR

WD’s usage guidance calls Red “SoHo” and Red Plus “SMB, also ZFS”. That’s clear.

See https://blocksandfiles.com/2020/06/...shingling-kerfuffle-gets-red-rebrand-bandaid/

It took them a good long while to get there, and they only got there because their use of SMR drives in a NAS line was exposed.

Still, this is the right way to go. Now there is a clear demarcation, and people know that all SMR are over in Red, and “Red Plus” is all CMR and will stay that way.

I trust that WD will stick to that promise, only because they are not suicidal. Every reviewer now knows what SMR is and how to test for it, and every channel would love to scoop the next stealth SMR. WD isn’t going to purposefully blow their own foot off. Again.
 

no_connection

Patron
Joined
Dec 15, 2013
Messages
480
IMO they shuold do like others and have a separate line or color for SMR. Red is still very much a NAS drive in 99% of forums and internet up to this point and while the media attention hopefully drive potential buyers to read up, but nothing really change if ppl just buy and replace what is on the sticker on the drive that came out of the device.

Funny thing is if WD would have launched an archive or low power "home/office" NAS drive with clear marketing they would have been pretty good off.

I hope IX keeps the momentum up with WD and can steer SMR into something that actually will work. We are going to need that high capacity drives sooner or later.
 

subhuman

Contributor
Joined
Nov 21, 2019
Messages
121
IMO they shuold do like others and have a separate line or color for SMR.
Like others?
This seems, to me, to be no different from what Seagate has done. Seagate Skyhawks are CMR, whereas Skyhawk Lites are SMR.

but nothing really change if ppl just buy and replace what is on the sticker on the drive that came out of the device.
If they replace an EFRX with an EFRX, no problem right? Every one I've seen has had the part number clearly on the sticker.
If people replace a failed part with a new part that has a different part number, that's on them to research if it's a viable replacement. At least, that's how I view it.
Keep in mind for your scenario it's someone taking the device apart and replacing it themselves. That's DIY, and with anything DIY the burden is on the person to do their research first.

if WD would have launched an archive or low power
That also could be very misleading or even downright false. With SMR, having to rewrite multiple sectors in order to change the data on a single sector may actually increase power consumption.
 
Last edited:

Yorick

Wizard
Joined
Nov 4, 2018
Messages
1,912
We are going to need that high capacity drives sooner or later.

SMR gives you a fixed increase in areal density, let's say on the order of 15%. It does this without needing to fundamentally change the recording technology, which is R&D intensive.

Seagate's HAMR and WD's MAMR - both "probably hopefully 2020" technologies - will drive increases in HD capacity, without using SMR. SMR can then be layered on top of that.

ZFS is not currently zone-aware. If it becomes zone-aware, it might actually be to better handle QLC NVMe SSDs, if (big if) WD's ZNS proposal is adopted by other players. See https://blocksandfiles.com/2020/06/...teradata-in-the-cloud-wd-zoned-ssds-and-more/. Zone-awareness could then also be used for HA-SMR and HM-SMR, it doesn't help DM-SMR one bit. DM-SMR is, arguably, never a good idea with anything but the most consumerish of consumer storage.

Note that "actually sequential resilver" is in code review, and will be supported by mirrors and draid, though not raidz. It can't verify checksum during resilver, but it does speed up small-block resilver on a mirror by 2x, and would presumably help SMR drives a lot. See https://github.com/openzfs/zfs/pull/10349
 

no_connection

Patron
Joined
Dec 15, 2013
Messages
480
If they replace an EFRX with an EFRX, no problem right?
Right, ppl like Patrick is clearly at fault here, not WD.

That also could be very misleading or even downright false. With SMR, having to rewrite multiple sectors in order to change the data on a single sector may actually increase power consumption.
As Yorick said, 15% more storage for same power draw is 15% less power since 99.999% of the time data will not be written. Extra power to write will be virtually insignificant when archiving data.
 

guermantes

Patron
Joined
Sep 27, 2017
Messages
213

subhuman

Contributor
Joined
Nov 21, 2019
Messages
121
Right, ppl like Patrick is clearly at fault here, not WD.
First, I never said no fault lies with WD.
What I am saying, is if a person replaces a failed part with a replacement that has a different part number, it's unreasonable to expect identical performance.
As Yorick said, 15% more storage for same power draw
I don't see where he made any mention of power draw. He said areal density, and the 15% was a hypothetical/guesstimate.
Every drive I've seen, and feel free to go through data sheets, draws more power during read/write than when idle.
SMR spends time doing housekeeping/garbage collection moving data from cache to SMR section of the disk. This is the higher "read/write" power level, not the lower idle power level. CMR does not need to do this. So, as I said previously, SMR may increase power consumption.
You're also revising your scenario. You initially said archival or SOHO, now you revised that to just archival.
 

no_connection

Patron
Joined
Dec 15, 2013
Messages
480
By archive I mean write data and not really change it, so the typical "scenario"is you fill it until it's full. And if you need to rewrite a lot and do a lot of moving files then ofc SMR is bad. Sure 15% was an estimate but if SMR had no benefit what so ever why is it even here?
I would still say that idle draw is the biggest time spend drawing power and less platters or smaller area for same data storage could still be beneficial.

What I am saying, is if a person replaces a failed part with a replacement that has a different part number, it's unreasonable to expect identical performance.
To me WD Red is still very rooted as NAS drive, and if someone that works with storage get caught out by it then I'm sure Bob from accounting that sees Red on the stick will just order another Red, and I would not blame him for not catching on that it's now called Red Plus for the same thing and old Red is now something else.
Part numbers often change with generation so using that as only indication can't really be generalized ether.
 

subhuman

Contributor
Joined
Nov 21, 2019
Messages
121
By archive I mean write data and not really change it, so the typical "scenario"is you fill it until it's full.
That's fine. But your original statement, the one I pointed out the possible flaws with, was for archival and "home/office NAS".
And none of this matters anyway. They didn't market them that way.
but if SMR had no benefit what so ever why is it even here?
Higher areal density. This does not always equate to lower power consumption. There can be a relation to the two, however they are not synonymous terms. You can only expect lower power usage from a SMR drive if the density crosses the threshold that allows you to use fewer heads (small power saving) or fewer platters (larger power savings).
Compare the WD30EFRX to the WD30EFAX: https://documents.westerndigital.co.../product-brief-western-digital-wd-red-hdd.pdf
The CMR WD30EFRX has both a lower idle and read/write power rating (2.7w and 4.1w respectively) than an SMR WD30EFAX (3.1w and 4.8w).
This brings us back to what I said in my first post in this thread: that marketing SMR as lower power may be misleading or downright false.
Again, note that HD manufacturers didn't market them as lower power. This is probably why.

To me WD Red is still very rooted as NAS drive, and if someone that works with storage get caught out by it then I'm sure Bob from accounting that sees Red on the stick will just order another Red, and I would not blame him for not catching on that it's now called Red Plus for the same thing and old Red is now something else.
Part numbers often change with generation so using that as only indication can't really be generalized ether.
And WD Red still may be viable for NAS use. NAS does not automatically mean "RAID" or "ZFS". Single-disk NAS units are being made and sold every day all over the world. They are very much the "archival use" you keep saying you're OK with seeing SMR marketed for.

Why is Bob from Accounting ordering your computer repair parts? Either Bob has the qualifications to do so, in which case yes it is his fault for making the mistake, or some other qualified person should be involved. If someone fills out a requisition for a "WD40EFRX" and Bob orders a "WD40EFAX" Bob is still at fault.
 

elorimer

Contributor
Joined
Aug 26, 2019
Messages
194
If someone fills out a requisition for a "WD40EFRX" and Bob orders a "WD40EFAX" Bob is still at fault.
Amazon led me down this path by telling me there was a "newer version" of the WDRed 4TB EFRX and I ordered it (for less $$). When it came it was an EFAX. Back it went, and now I know (thanks to the folks here).
 

Tigersharke

BOfH in User's clothing
Administrator
Moderator
Joined
May 18, 2016
Messages
890
Amazon led me down this path by telling me there was a "newer version" of the WDRed 4TB EFRX and I ordered it (for less $$). When it came it was an EFAX. Back it went, and now I know (thanks to the folks here).
Be very careful of Amazon and what is effectively a sales pitch. Surely it is some sort of algorithm so the code has no clue when it is wrong.
 
Top