.

[Concillian]

It begs the question, though, why HDD manufacturers all cite figures for load/unload cycles if they're not that important? Esp. in WD's case, why make a drive that cycles load and unload so frequently that you're all but guaranteed to surpass the 300K rating? Either their spec sheet is wrong, or the drive's design is off. Pick one.

Because customers in many industries don't necessarily update their specifications for obsolete ones.

They had a problem 10 years ago with this issue, that spec stays on their spec sheet FOREVER. Regardless as to whether the design has changed to obsolete that as a spec.

People were once sensitive to the issue, and people have long memories on this kind of thing. So it remains a spec.

 

[tidy]

.

 

[bononos]

No they didn't, my WD20EARX was still 8 seconds.

Seagate does.

The wdidle3 utility says that the factory setting for head parking on a 1Tb earx that I checked was 30s.
I remembered reading a seagate whitepaper last year which said that the standby timeout was 4mins (I know there are Seagates which have 8s head parking timeout).

 

[tidy]

.

 

[Carson Dyle]

Why would moving the heads to a parked position cause any more wear that moving them to do a read or a write? And how many thousands of times a minute does that happen in normal use of a hard drive?

 

[tidy]

.

 

[Carson Dyle]

I still don't understand why you would think there's any "wear" involved. Do you think they allow the heads to come in contact with something? Why would they have to?

 

[tidy]

.

 

[code65536]

Because customers in many industries don't necessarily update their specifications for obsolete ones.

They had a problem 10 years ago with this issue, that spec stays on their spec sheet FOREVER. Regardless as to whether the design has changed to obsolete that as a spec.

People were once sensitive to the issue, and people have long memories on this kind of thing. So it remains a spec.

It's not like they've never changed it, though. 3.5" drives were only recently changed to 300K or 600K; it wasn't that many years ago that they were a whole order of magnitude less (like 60K, IIRC).

And at least in the case of Seagate, they went through the trouble of doubling the cycle rating for their NAS model to 600K from 300K (I haven't reviewed WD's specs, so maybe they do something like that too).

Anyway, even if the evidence suggests that the cycle count doesn't matter, it's hard to silence that anxious voice in the back of the head because of what these spec sheets say.

 

[Zxian]

Something comes into contact with something. I'm not saying it's the actual heads themselves. I don't know - I'm the one asking the questions.

That's precisely my (and Carson Dyle's) point. Why does it matter?

Suppose that everyone was right, and that the LLC count really did correlate to the 300,000 load/unload cycles specified by the datasheet. We'd be seeing far (and I mean far) more reports of failed Green drives than we do. The data just doesn't add up.

If you don't like it for performance reasons, that's another matter, but worrying about a mechanism that may or may not cause problems is strange to me.

 

[Concillian]

Anyway, even if the evidence suggests that the cycle count doesn't matter, it's hard to silence that anxious voice in the back of the head because of what these spec sheets say.

Is 300k a problem? I was kinda assuming this number of cycles wasn't an issue, but you seem to be pretty concerned.
That's cycling every 10 minutes day and night for 5 years. Does your computer never sleep, but every 5 minutes accesses some data? This is a very strange usage profile that would have the drive spinning up and down so frequently.

Normal users at least turn the computer off while they sleep and have large idle periods while they eat, drink, watch sports, socialize, etc...

I think you're underestimating just how many cycles 300,000 is. If you do indeed have a 24/7 usage profile that spins a drive up and down every 5 minutes or less, then I think you just plan on replacing drives more often. However, a usage profile where the computer is doing nothing enough that the drive spins down, but something frequently enough for 300,000 cycles to be a concern over a 5 year lifespan is a crazy weird usage profile.

 

[code65536]

Is 300k a problem? That's every 10 minutes day and night for 5 years. Does your drive cycle even close to every 10 minutes day and night seven days a week? If you power it down when you sleep, then that moves it up to parking every 6 minutes for 5 years? I think it's pretty difficult to get over about 50 cycles a day, which is like 15 years for 300k.

Spin up and spin down take a certain amount of time themselves. I think this is the kind of thing that if you have a usage profile where 300k is a concern, you plan on buying a replacement drive more frequently. This is similar to a specific write heavy SSD usage profile... you just have to plan on replacing the drives more often (or buy specialized expensive high write cycle SSDs.)

Load/unload cycles are not spinup/down cycles. Unloading the head is when the head moves off the platter (often with an audible click). And with some drives, it can happen every few seconds, not minutes. I have a drive running 24/7 that racked up around 600K cycles in less than a year of service. And the official spec sheets for a lot of drives say that they're rated for either 300K or 600K of these sorts of cycles.

 

[bononos]

Load/unload cycles are not spinup/down cycles. Unloading the head is when the head moves off the platter (often with an audible click)........

I'm pretty sure they are. The heads are parked off platter, the platters stop spinning. It corresponds to the standby state on the Seagates/WD.
And its just common sense, why would the platters continue to spin when the heads not on the platter?

 

[bononos]

That's really odd, because wdidle3 /R definitely showed 8 seconds for my 2TB EARX.
......
I was thinking of buying a WD 'red' drive, but there were complaints about head parking even in those "NAS" drives.

Yes its odd because this guy says that the EZRX timeout for head unloading is 8s as well.

Theres a utility for modifying the NAS reds in this WD forum thread. The WD Red SMART load/Unload utility doesn't seem to be the same as the wdidle3 one.

 

[Carson Dyle]

I'm pretty sure they are. The heads are parked off platter, the platters stop spinning. It corresponds to the standby state on the Seagates/WD.
And its just common sense, why would the platters continue to spin when the heads not on the platter?

Because the time and energy required to load and unload the heads is just a tiny fraction of that required to spin up the platters. The latter would have a potentially huge effect on performance. If we're talking about intervals as little as 8 seconds, there's no way the drives are spun down.

 

[tidy]

.

 

[bononos]

Because the time and energy required to load and unload the heads is just a tiny fraction of that required to spin up the platters. The latter would have a potentially huge effect on performance. If we're talking about intervals as little as 8 seconds, there's no way the drives are spun down.

Yes but its odd that the documentation says that the platters aren't spinning.

http://www.seagate.com/staticfiles/support/docs/manual/desktop/Barracuda 7200.14/100686584.pdf

http://www.seagate.com/staticfiles/support/disc/manuals/desktop/Barracuda Green/100649225b.pdf

.... In Standby mode, the drive buffer is enabled, the heads are parked and the spindle is at rest. The drive accepts all commands and returns to Active mode any time disk access is necessary.

Power modes Heads Spindle Buffer
Active Tracking Rotating Enabled
Idle Tracking Rotating Enabled
Standby Parked Stopped Enabled
Sleep Parked Stopped Disabled

If you look at the spec sheets, the sleep and standby modes consumes almost the same amount of power suggesting that the platters have stopped.
http://www.wdc.com/wdproducts/library/SpecSheet/ENG/2879-771442.pdf

 

[tidy]

.

 

[UaVaj]

WD is that stupid. parking the head does not save power. spinning down does.

will not touch a green unless there is a laptop requiring a 3.5" internal hard drive.

 

[code65536]

standby mode

But that's not what we're talking about.

Hard drives have different states. I don't know what the technical names for them are (they probably vary from manufacturer to manufacturer anyway), but...
* active: platters spinning, heads poised over the platter
* parked/unloaded: platters spinning, heads moved off the platter
* sleeping: platters no longer spinning, heads moved off the platter

This thread is about that second state, where the platter keeps spinning while the head is parked. Head parking does not necessarily mean that the platters have spun down. On the other hand, the platters spinning down does necessarily mean the heads have parked (the spinning motion of the platter creates a flow of air that acts as the cushion separating the head from the platter, without which the head would crash into the platter).

Spinning down and spinning up a platter is time-consuming and can take many seconds (even up to a quarter of a minute or more), whereas parking/unloading and unparking/loading the head takes a small fraction of a second (though it's enough to substantially impact access latency vs. the case where the head isn't parked).

With WD green drives (and many notebook drives), the head parks very frequently. Often once every few seconds. If you've ever heard a drive clicking every few seconds (not as loud or frequent as the click-of-death, which is a different sort of click), that's the head being frequently parked/unloaded.

Unloading/parking used to be done mostly as a preemptive safety precaution on laptop hard drives. By unloading the head when you're idle, it reduces the amount of time the head is floating above the platter, thus reducing the probability that the head is in a position to crash into and damage the platter in the case of a sudden jolt. Some drives also have motion detectors that will essentially do an emergency park of the head if it thinks that it's falling. You don't see this on 3.5" drives because desktops and servers are not subject to the jostles of a mobile computer.

As someone has already pointed out, parking doesn't really save much in the way of energy because the largest consumers of power in a HDD--the spindle motor and the chipset--are still working. Spec sheets list power for load (when the head is busy writing or reading), idle (when the drive is spinning, but the head isn't doing anything--it may or may not be parked), and standby/sleep, where the drive is spun down. They have never released separate numbers for "idle, head parked" and "idle, head still floating above platter", probably because any difference between them would be lost in the rounding.

Which is why the aggressive parking on the WD Greens is strange, because, unless the drives are being used as portable externals where maybe one might have to worry about laptop-style movement, frequent parking just doesn't make sense because it doesn't save any power and just causes unnecessary amount of extra head movement.

 

[Carson Dyle]

Which is why the aggressive parking on the WD Greens is strange, because, unless the drives are being used as portable externals, frequent parking just doesn't make sense because it doesn't save any power and just causes unnecessary amount of extra head movement.

These are the same drives that they package and sell as externals, so maybe it does make some sense.

 

[bononos]

But that's not what we're talking about.

Hard drives have different states. I don't know what the technical names for them are (they probably vary from manufacturer to manufacturer anyway), but...
* active: platters spinning, heads poised over the platter
* parked/unloaded: platters spinning, heads moved off the platter
* sleeping: platters no longer spinning, heads moved off the platter

This thread is about that second state, where the platter keeps spinning while the head is parked. Head parking does not necessarily mean that the platters have spun down.......

Thats was what I was getting in my previous post, the portion about the idle state I quoted from the seagate document says that the head are tracking (not parked) and only the standby and sleep state shows head parking.

Maybe WD drives don't spin down completely and start from a dead stop, they could just spin much slower in the standby state.

 

[tweakboy]

No.