[SOLVED] WD Red NasWare 3.0 data protection in case of power loss?

[Guest_1]

WD's page on NasWare 3.0 states the following:

NASware further improves drive reliability and protects your data in the event of a power loss or disruption.

COMMAND COMPLETION
To further improve reliability, NASware minimizes the data corruption or loss in the event of an unexpected power loss by completing the command in process before shutting down.

I'm not sure how firmware alone could accomplish this. Do WD Reds have super caps or some other circuitry to ensure command completion after power loss? Do the WD Reds actually protect your data in case of a power loss or is this just a fancy way of saying they'll expect the RAID controller to protect your data?

I live an area where unexpected power loss is quite frequent, so this feature is extremely important to me but I need to know how it works to make sure it's not just marketing BS.

I'm going to put a 4TB WD Red drive in a desktop PC which remains on 24/7. The WD Red will be the only drive in the system so it obviously will not be in a RAID configuration. My reason for choosing the WD Red is that it's the only drive rated for 24/7 operation other than the Purple.

 

[kanewolf]

If frequent power loss is common, then you need a UPS to protect your equipment.
My guess it that the drive does have a supercap. But I wouldn't rely on that time after time.

 

[fzabkar]

If you can provide a model number, I could examine the PCB for you.

BTW, many "WD" drives (eg WD80EFZX) are rebadged HGST models. These HGST drives have their own firmware. This begs the question, if "NASware" is a WD firmware feature, then can it really be present in a HGST firmware architecture? Does the equivalent HGST model also make claims about NASware?

There are photos of some WD/HGST models here:

www.hddoracle.com/viewtopic.php?f=21&t=2549
https://hardwarecanucks.com/storage/western-digital-red-8tb-helium-hdd-review/

There are several PCBs which have banks of capacitors which are only sparsely populated.

Here is one example:
www.hddoracle.com/download/file.php?id=8606&mode=view

Notice that there is one capacitor array (near the RH cutout) which can accommodate 9 capacitors, but which is only populated with a single capacitor. The array above it only populates 2 out of 7. I don't know the implication of these omissions, but it would make me suspicious of any claim regarding power loss protection. I would ask someone at WD, preferably an engineer, for clarification.

Here is a second example:
http://images.hardwarecanucks.com/image/akg/Storage/WD_RED_8TB/board.jpg

Above the RH cutout is a capacitor array set aside for 3 solid tantalum capacitors, but only one 47uF capacitor is populated. The previous example has the same capacitor. Is this a power loss protection capacitor? If so, does it allow the drive to write to the platters, or does it merely allow the drive to retract its heads before the voice coil loses power completely?

This is how a typical SSD implements power loss protection:
http://www.hddoracle.com/viewtopic.php?f=59&t=231&p=21248#p21248

 

[Guest_1]

If you can provide a model number, I could examine the PCB for you.

I don't know the implication of these omissions, but it would make me suspicious of any claim regarding power loss protection. I would ask someone at WD, preferably an engineer, for clarification.

The model number of the drive I'm considering is WD40EFRX .

It's pretty hard to get a hold of an average WD phone representative who knows less about tech than the average highschooler, it would be pretty near impossible to get a hold of a WD engineer unless you have sources inside the industry.

 

[fzabkar]

Here is the PCB from a WD40EFRX-68WT0N0:
https://www.hdd-parts.com/14060217.html

There is a bank of 4 capacitors near the motor connector, but I don't recognise anything that is related to power loss protection.

The typical power loss protection scenario in older drives uses the spindle motor as a generator which provides the energy for an emergency retract of the headstack. Power loss protection in an SSD utilises the stored energy in a bank of capacitors to power the switchmode regulators so that data can be flushed to NAND array. There doesn't appear to be anything like this on this particular PCB.

 

[Guest_1]

Here is the PCB from a WD40EFRX-68WT0N0:
https://www.hdd-parts.com/14060217.html

There is a bank of 4 capacitors near the motor connector, but I don't recognise anything that is related to power loss protection.

The typical power loss protection scenario in older drives uses the spindle motor as a generator which provides the energy for an emergency retract of the headstack. Power loss protection in an SSD utilises the stored energy in a bank of capacitors to power the switchmode regulators so that data can be flushed to NAND array. There doesn't appear to be anything like this on this particular PCB.

I don't know where the motor connector is supposed to be so I can't see those capacitors. Do you know the capacity of these capacitors?

If this HDD doesn't use supercapacitors to flush it's cache then do you have any alternate theories on how the firmware 'command completion' on power loss is supposed to work?

 

[fzabkar]

The motor connector comprises 4 terminals at the bottom left edge of the PCB, immediately to the right of the screw hole. There doesn't appear to be anything "super" about the capacitors on the 45 degree edge of the PCB, so I have no idea how WD can make their claim. If they are relying on the residual capacitance in each of the switchmode regulators, then that wouldn't give the drive much time to do anything, at least not reliably.

In fact, I have no faith in any claim by WD, at least not those that originate in their marketing department.

Here is something new to watch out for:

A new twist to IntelliPower?
https://forum.hddguru.com/viewtopic.php?f=3&t=38333

If you purchase a "5400 RPM class" HDD, you may find that it actually spins at 7200 RPM, despite reporting a 5400 RPM spin speed to the host. This begs the question, is it a "5400 RPM class" HDD because it fails the acceptable performance expectations of a 7200 RPM HDD?

 

[Guest_1]

Thanks for linking the HDDGuru thread. I think I'll post this question there , maybe someone over at a forum of HDD experts knows exactly what's up.

 

[popatim]

LOL - that is him too.

AFAIK, Nasware's Data Protection merely completes the current command before shutting down. I suspect Fzabkar is correct that the spin motor is used as emergency power to finish the current sector write, if one is in progress, and park the head and shutting off. I don't see this as any 'data protection' sinc eyou file will only be partially written. Maybe they mean it in the sense of 'Hey we parked the head so that it doesn't crash to the platters and take out all your data"... Typical marketing mumbo jumbo

 

[fzabkar]

This is interesting:

WD Gold EX1000M Datacenter Capacity HDD (WD2005FBYZ / WD1005FBYZ):
http://wfcache.advantech.com/www/certified-peripherals/documents/96hd1tb-st-wd7ke1_datasheet.pdf

Power Loss Data Protection - Upon power loss, the drive utilizes stored spindle energy to back up the HDD cache to on-board flash. This allows deeper write queues which boosts performance, while minimizing data loss/corruption such as write splices that can occur during unexpected power losses.

This thread has a photo of the PCB:
https://wdmarvel.com/forum/ru/viewtopic.php?t=422&start=190

There is a JMicron JMF608u2 NAND flash controller:
shop.maxxxware.de/external/datasheets_public/renice/jmicron/jmf608.pdf

This would suggest that the PCB is from an SSHD???

However, there is a link to this other PCB (2060-800032-004):

This second PCB has two serial SPI flash chips at U12 and U15. U12 traditionally stores the firmware and "adaptive" data, so I expect that U15 could be the backup flash memory for the RAM cache. Its capacity appears to be 8 Mbits (1 megabyte). There is a large bank of 11 capacitors nearby.

 

[popatim]

Well 1mb isn't going to backup the drives volatile cache. 128MB+ just wont fit even with compression.
They have to be continuing to write the buffer contents even as the drive spins down in that case; which would be cool. But, if you were in the middle of transferring a large file, it wouldn't finish and you'd still have a corrupted file on the HDD.
I'm just not buying into any data protection claims. LoL

 

[fzabkar]

The 4TB PCB ( WD40EFRX-68WT0N0) pictured earlier has no additional SPI flash, nor is there a similar bank of capacitors, so I'm wondering whether WD implements the power-loss-protection feature in one special model and then applies a sweeping generalisation to all its NasWare models. In any case, something isn't right. I would like to see the original web page where this feature is advertised. I'm still open for someone to tell me how it's all supposed to work. Perhaps "Command Completion" is not the same thing as "power loss protection" ???

BTW, I'd also like to know how WD defines "RPM class". Maybe they have their own dictionary or their own special rev counter?

 

[popatim]

I had heard thru the rumor mill that anything that couldn't meet 7200 rpm spec was badged as a 5400 even if it hit 7000.
My theory is that if they run out of 5400 rpm drives and need to get an order out, they just use the 7200 rpm stock. LoL

 

[fzabkar]

This web page would suggest that "Command Completion" is a consistent feature of NasWare, which is "built into every WD Red hard drive":

https://shop.westerndigital.com/solutions/network-attached-storage/nasware

Here is a WD80EFAX-68KNBN0 "8TB Red" which is a rebadged HGST model:
http://forum.hddguru.com/download/file.php?id=18635&mode=view (PCB component side)
http://forum.hddguru.com/download/file.php?id=18634&mode=view (HDA)
http://forum.hddguru.com/download/file.php?id=18633&mode=view (PCB bottom side)
http://forum.hddguru.com/download/file.php?id=18632&mode=view (Drive label)

Its label proclaims "NASware 3.0". The PCB has two 47uF solid tantalum capacitors and two BGA SPI flash memories, at least one of which must contain the firmware.

 

[Guest_1]

LOL - that is him too.

I know.

U12 traditionally stores the firmware and "adaptive" data, so I expect that U15 could be the backup flash memory for the RAM cache.

Isn't firmware usually stored on the platters?

I would like to see the original web page where this feature is advertised.

There's a link to the page in my very first post.

It's amazing to me how WD and other major consumer HDD manufacturers intentionally obscure detailed specs of their products. It just shows they either have something to hide or their marketing department is putting out straight up lies.

 

[fzabkar]

Here is another one (6TB WD Re WD6001FSYZ-01SS7B1):
http://www.hddoracle.com/viewtopic.php?f=3&t=2762

This "enterprise class" drive incorporates WD's newer ARCH7 firmware architecture. Location U12 is occupied by the firmware as in earlier models, but there are an additional 6 x Winbond 8Mbit SPI flash memories at U15 - U20. These 6 "ROMs" were mostly empty when the owner dumped them. At the time we were all puzzled as to the function of these ROMs, but it now appears that they may be involved in power loss protection. In fact there is a bank of 11 capacitors nearby, and these appear to be charged via a diode in a boost converter configuration.

The datasheet confirms that Power Loss Data Protection is present in capacities of "5 TB and above":

https://www.elfadistrelec.fi/Web/Downloads/-t/ds/WD-RE_SATA_2-6TB-eng-tds.pdf

Here is the PCB from the 4TB drive on the same datasheet (WD4000FYYZ):

It has a single ROM at U12 and no capacitor bank, ie there are no power loss protection components.