I saw this option in the Xfce power management to never spin down hard disks. What exactly does that entail? Does the hard disk spin at a lower speed? I read that "spinning down" shortens the life of your drive, but wouldn't less disk motion mean less wear and tear? Can someone please clarify?

When spun down the HDD is not rotating. You can hear that as the disk is spun down.

Some (all?) bearing types suffer most wear during stopping and starting, hence the belief that spinning dow disks shortens their lives.

its a power saving option. it takes the HD down to not spinning and using power. as in a laptop to save your battery life.

usually first you turn off the monitor first say after 10 minutes of non usage as an example
then after 10 minutes more you might set the HD to spin down.

as I said on a portable device it has more of an affect of saving battery life.

on a desktop its a mute point but could save a few pennies a year in hydro

What it entails is the software never spinning down the hard disk (while there is power, obviously....if the power to the hard disk fails, then it still spins down, irrespective of this setting).

The majority of hard disks spin at the speed that they spin at (one speed only), but some hard disks have the ability to vary the spin speed, but that is under the control of the hard disk and not XFCE. Typically, manufacturers have some kind of brand name for this ability to vary the spin speed. So, for example, Western Digital has 'Intelllipower' on some of its drives where the drive detects that it is not busy and goes down to the lower spin speed (5400 down from 7200 rpm, in this case, although the numbers can vary in other cases; HGST's 'Coolspin' is probably similar in principle, but the numbers for HGST are probably different).

Quite often you'll see this sort of feature on 'Green' drives, but not exclusively. Sometimes 'Raid' or 'Home Cinema' drives have this feature, too.

Wear on what? You have to understand mechanical hard disk drives. What keeps the head off the disk is that aerodynamic forces keep the head 'flying' (it flies at some fraction of the width of a human hair above the magnetic surface). When the head stays above the disk surface by an adequate margin, physical contact doesn't occur, which reduces wear in that area. When it doesn't, wear is distinctly more possible. Now, this isn't usually as bad as it sounds, because the wear is constrained to a non-active area, when it is controlled, but the head still ends up with magnetic material on it, and that can cause problems, and the head itself can wear, although that isn't usually a problem and there can be magnetic material inside the disk and given that it is abrasive, that can be bad news.

So, landing the head on the disk has the potential to increase wear, even though the hard disk is off for more of the time.

I was under the impression that when power is cut to the head's voice coil it moves back to a position where it is held above the surface of the disk and can't touch it in any way? Much the same as when old stepper-motor hard drives used to have to be told to park the head before switching off.

You're talking about a head crash. If the head even slightly touches the disk, a head crash will occur and the drive will be permanently damaged. This is why so many old laptops suffered hard disk failures. This is not a problem with modern hard drives because of a feature called active hard-drive protection which immediately moves the head away from the disk if any sudden movements occur.
@salasi I was talking about how all of that spinning might wear down the axle mechanism over time.

Alright, but how would spinning down a hard disk shorten down its lifespan? And how much? Shoud I enable hard disk spindowns?

Maybe there are no definitive studies providing the information to answer that question (I looked and did not find).

HDD motor and axle bearings used to be ball bearings. In motion, a film a lubricant is forces between the balls and the tracks so there is no metal-to-metal contact. During startup the fluid dynamics required to pump lubricant between ball and track do not exist. Avoiding metal-to-metal contact, depends on the residual film of lubricant which is held in place by "wetting". Those bearings were affected by repeated startups, especially as they got older and the lubricant decayed.

I vaguely remember reading somewhere that bearing technology has developed and this is effectively no longer an issue. Ball-bearings have been replaced by fluid bearings (http://en.wikipedia.org/wiki/Fluid_bearing). They are cheaper, longer lasting and quieter than ball bearings. Unfortunately I found nothing to say that they are less susceptible to startup wear

There's another way to consider this. What are the benefits from spinning down: savings in electricity and reduction in consequent waste heat (AC costs to remove it etc.); less noise ... What are the costs: unquantified risk of reduced lifetime. To judge the significance of the reduced lifetime, you could factor in the proportion of HDDs that die because of worn bearings as compared with recording medium failure (head crashes and bad blocks), because of electronic failure, motor failure etc.

I dont believe that the WD IntelliPower drives do change rotation speed-

http://www.xbitlabs.com/articles/sto...roundup_2.html

http://techreport.com/review/15769/w...een-hard-drive

I dont mind the WD 'Green Power' drives, I own one myself and so does my housemate. But I'm still unimpressed with the marketing BS they did with the GP drives and 'IntelliPower'.

I would guess that Hitachi GSTC is doing the same trick, i.e. not listing the rotation speed because of marketing reasons. So far I havent seen any reference to the 'Coolspin' drives changing rotation speed.

WD-RE4-GP, WD AV-GP and WD Red drives are just standard WD Green Power drives with different firmware as far as I know.

Wow. I'd just seen the data sheets with 5400 - 7200 on them, and to use that description for a drive with a single rotational speed is so close to lying, I wonder why they would do that.

I usually turn off my computer after each use, which obviously spins them down, and my hard drives have given me 5-8 years of life, so I really doesn't think it matters with my situation. I'm just going to leave that setting alone.

Hard-drives before Winchester would move the heads to outside of the platter where an arm would catch the head carriers. These platters were 12" to 18" across, and often had removable disk packs.
Winchester drives introduced drives that moved the heads to the near the spindle and landed the heads on the surface. They touch down in the landing zone which has no data. Winchester drives do not allow removing the platter, which allows much denser data because the data is always read and written with the same head.

With faster and variable rotation speeds, there has been some talk about drives that may once again catch the heads instead of landing them, but finding out may be difficult as manufacturers don't want to disclose details.
One problem is that when launching the heads from a catcher they bounce, and may contact the surface with more damage than landing them would have.

All drives I have encountered (since 1970's) have a protection mechanism that yanks the heads off the data area when power fails. It puts the heads where a commanded spin-down would have.
A commanded spin-down moves the heads to the landing zone and powers off the drive motor.
Normal computer shutdown also does a commanded spin-down of every drive before the power is turned off.
In all these cases the heads end up landing in the landing zone. They land when the platter spins down to a speed where the induced air flow will no longer support the heads.

Ad-hoc rules.
1. It is a trade-off between power used, drive lifetime, and access delay.
2. If access delay is critical (as in a server), do not spin down the drives. You will pay for this by using max power all the time. But even on a server, there may be some drives that do not need to be spinning all the time.
3. If power is critical (as in a laptop), then spin down the drive after a reasonable time, such as 15 minutes. This could be as short as 5 minutes.
4. If lifetime is critical, the answer is not as easy.
5. Spinning down a drive every 10 minutes is not recommended for a desktop machine. The minimum idle time to spin-down should be 15 minutes or more, and only when it will stay idle for a hour. This will vary if you have a drive that can tolerate more landings.
6. If a drive will be idle for more than 1 hour (such as machines that are used infrequently but left running), it is probably acceptable to spin down that drive after 30 minutes of idle. Maybe longer.
7. If a drive will be idle for more than 4 hours (such as machines that are left running overnight), it is good to spin down that drive after 60 to 90 minutes of idle.
8. Your exact times may vary from my suggestions.
9. Solid-state drives have much different spin-down costs.
10. Do not spin-down the drives of a RAID-setup independently. RAID systems with idle periods should consider changing to a non-RAID setup with independent drives, or a partial RAID setup. Starting all drives of a RAID system at the same time will stress the power supplies which will shorten the lifetime, and staggered starts have increased access delay.

I go by these questions:

Is it portable if yes then SSD, if no then platter. If it's old and portable I buy CF to IDE and use CF cards in them.

OK, here's my ... Hard drives, as mentioned, had a special landing area for the heads when the disk was powered down, this area allowed the heads to rest safely without any damage or data loss. I believe the parking area had a treated surface to prevent damage. As time went on, this area deteriorated and you also had the main bearings for the spinning platters starting to dry out. This is generally fine as the disks keep spinning though the bearings will eventually start getting noisy. (I'm thinking of a 30 disk Sun SA100 array in particular that I encountered where the disks had been spinning for probably 10 years! It sounded like a saw mill!!)

Spinning a disk down was OK but trying to spin it back up was like trying to move your car with the disk brakes locked; the heads were gripping the platters and preventing it spinning up. This is where the "bean shake" comes in where you had to shake the disk in the same plane as the platters to break the stiction preventing it spinning. This was fun with the SSA 100when replacing a faulty disk as the smallest number of disks which required to be spun down to do this was 10, one disk trays worth! You invariably had a couple which wouldn't start.

Modern disks have ramp loaded heads which cures the stiction problem.

As far as I know the disk speed is constant; 5400, 7200, 10k, 15k rpm. To use disks of different speeds in a RAID is not advised. What happens is that instead of sequential sectors being read, the read access of the faster drives is compromised, being cut back to that of the slower drive. The spindle speed isn't reduced but the disk's cache has to compensate. Where similar disks could stream data sector by sector, mismatched disks would see sector reads being done on alternate rotations, etc, with the disk cache having to regulate the data flow. (Does that make sense?)

Ah.. These were the days

Play Bonny!