On Fri, 05 Feb 2016 10:16:21 +0000, Java Jive wrote:

On Fri, 05 Feb 2016 02:20:20 GMT, Johnny B Good
wrote:

On Wed, 03 Feb 2016 20:05:25 +0000, RJH wrote:

On 03/02/2016 04:47, Johnny B Good wrote:
On Sun, 24 Jan 2016 09:48:15 +0000, RJH wrote:

There's no hard and fast rule regarding the use of spin down power
saving in a SoHo or home NAS box but, unless you're really only
making infrequent use of the NAS, it's always best to avoid lots of
spin up events per day (most home desktop PCs are typically power
cycled just one or two times a day which keeps the spin up event
count nice and low, assuming that distraction known as spin down
power saving in the OS has been completely disabled in order to
preserve the operator's sanity).

I don't think it's asking too much of any user to wait for a HD in a PC
or NAS to spin up when it's not been accessed for a long time. One just
gets used to it.

Or, as I presume they must say in MSFT's marketing division, "One just
gets habituated to the situation." :-)


It's worth keeping in mind that this is a *power saving* feature
(in
reality, an energy consumption saving strategy) with no thought to
whatever consequences there might be in regard of the drive's
reliability. Seagate must be the only drive manufacturer stupid
enough to confuse power saving with temperature reduction if their
FreeAgent 'specials' were anything to go by.

It is certainly true that saving power has to be considered along with
product life. The world is full of examples of electrical and
electronic products that are designed to run 24/7 - fridges and
routers, for example - and particularly with the latter switching them
off overnight may lead to premature failure, which, when the economic,
environmental, and energetic 'costs' of manufacture and disposal of the
products are considered, may be less economic and less ecological, than
just leaving them on 24/7 as they were designed to run.

However, I suspect that is not true of HDs, which were designed to spin
up and spin down to save energy.

Actually, they weren't designed for that use. The power saving spin down
is a feature added to laptop drives that then became a standard add-on
option in the larger desktop drives shortly afterwards.

In the case of laptop HDDs, such energy saving strategies do seem to
work without the same detrimental effects witnessed in their larger
desktop cousins (I'm thinking of the ten year old WD *IDE* laptop drives
also afflicted with the same 8 second head unload time out discovered in
the desktop green models circa 5 years ago with head unload cycle figures
of 3 and 5 million, a value that's a magnitude larger than the quoted
300,000 lifetime rating for those green models - durability in the face
of head unloading 'wear and tear' doesn't seem to scale very well in the
larger desktop models).

In fact, in the case of laptop usage, this 'insanely' short 8 seconds
head unload time out makes quite a lot of sense in that it vastly
increases the chance that the heads will be safely parked if the lid is
accidentally slammed down too hard or the laptop dropped too hard onto a
desk or it gets kicked off the desk onto a hard floor. Also, spin down
power saving in this scenario is more likely to offer a net benefit on
lifetime, not only for the drive itself but also on the rest of the
laptop's components.

The same is hardly true in the case of desktop drives and it's a pity WD
didn't rethink the 8 seconds default time out on head unloading,
especially as it only had a rather modest 300,000 cycles rating on the
Greens (600,000 for the REDs). Although the extra access delay is only a
matter of half a second or so, the 3 to 4 hundred milliwatt power saving
is a rather questionable benefit (unless you're looking for 'Kudos' from
the dumb assed reviewers for being just that little bit more "Greener"
than the 'Competition').


Spinning down a modern HDD typically reduces power consumption by
around
7 to 10 watts per drive as observed in the energy consumed at the
mains socket. Each watt year of energy consumed equates to about a
quid's worth on the annual electricity bill. That represents 8.766
KWH units of electricity used per year. You can check your actual
unit costs and calculate a more exact annual cost per watt's worth
of 24/7 consumption.

If you're running the NAS 24/7 and just using spin down power
saving to
minimise its running expenses, you can estimate just how much of a
saving this contributes by calculating the hours of spin down
'sleep' time each drive enjoys per day. For example, a pair of
drives allowed to 'sleep' overnight may get anywhere from 8 to 16
hours of repose per day, depending on how often you access the files
on the NAS box and the timeout period you've selected before the
drives spin down.

For arguments sake, I'll assume an average of 12 hours per day of
spin
down sleep for both drives and an effective energy saving at the
socket of 10 watts each, 20 watts in total making for a saving of
240 watt hours per day. this represents a total of 87.66 units of
electrical consumption saved over the year. Assuming 15p per unit,
this would represent £13.15 savings on the yearly electricity bill.

This doesn't strike me as a worthy enough saving to place the
drives
under the additional thermal cycling stresses introduced by such a
power saving strategy.

[snip]

I know because, barring silly manufacturing defects or system design
errors that expose the silicon to electrical stresses beyond their
design limits, thermal expansion/contraction introduces mechanical
cycling fatigue induced stresses on the silicon die as well as in
circuit board through plated holes.

[snip more of same]

Frankly, IME this is ********. I cannot recall a single HD failure in
the electronic PCB, every single one I've ever owned has failed due to
bad sectors developing on the platters. How many drives have you had
fail in the way that you claim? I'd be surprised even at a single one.

Sadly, googling "hdd spin down life rating figures" and variations of
this phrase in the hopes of being taken directly to a manufacturer's
spec sheet (or an article with such links) only produced discussions in
various web fora on the pros and cons of spin down power saving where
the only 'nuggets' were ill informed opinion best described as "Pearls
of Wiz-
Dumb"

Quite, so why are you helping to create/perpetuating yet another urban
myth? The facts on this particular topic are that there are no facts,

It's not that there are *no* facts, just that it's hard to track down
any published figures in this regard. There's absolutely no doubt that
temperature cycling is detrimental to the life ratings of all such
electro mechanical systems, it's simply a question of just how important
it is to a drive's useful life which, until recently could easily exceed
the 4 to 5 years it takes to "Outlive its usefulness" until the
manufacturers fine honed their "F1 GP race car design" approach to
minimise the expense of such 'over-engineering' which lead to the
"Outlive its usefulness" effect in the first place.

so you have no business peddling one viewpoint over another,
particularly when you're going against most users' experience,
including, I would guess, even your own.

I can't speak for others' experience but I can certainly remove the
guesswork from your presumption about my own which reinforces the idea
that the drives in a NAS box operating full time are generally best left
spinning 24/7 unless you have a very well defined usage pattern that
allows the time out period to be tuned to minimise the number of spin
down cycles per day.

As for the business of "peddling one viewpoint over another" as you put
it, on that basis neither do you. In fact neither of us can lay claim as
to which is the best strategy to use with any great authority but we're
certainly both entitled to offer our opinions (preferably, reasoned
opinions).


Now, I have looked at that - and changed the spin-down triggers to 1
hour.

That seems a more reasonable compromise between MSFT's choice of 20
minutes and my own of 2 or 3 hours.

My PC drives spin down after 5 minutes when running off mains power, the
laptops after 3 minutes when running off the battery. From memory I
think the NASes are the same as the PCs running off mains. I find the
resulting usability and reliability both perfectly acceptable.

I'm afraid a 5 minute time out would drive me crazy with its 10 to 12
seconds access delay almost every time I need to read from or write data
to disk. As I previously alluded, disabling spin down power saving wasn't
the only consideration in finding a balance between reliability and
'economy of energy consumption'. The other was in regard of my mental
health which helped decide the question as to whether or not it was worth
risking reliability for a modest saving in energy costs.

I had become rather habituated to 'instant gratification' with all of my
home built external usb connected drives so it came as quite a shock to
experience such delays perpetrated by a "Seagate Special" with the 10 or
15 minute spin down hard programmed into the drive controller's firmware
used in a Medion re-badged 500GB FreeAgent drive.

Browsing the various web fora where this 'annoyance' was discussed at
length failed to elicit a solution so, in the end, when the opportunity
arose, I was able to repurpose the drive as a replacement in a Vista box
where not only would it never spin down ever again, even better, it would
enjoy much better cooling and therefore stay safely below the 60 deg C
temperature limit it had managed to hit despite my precaution of using a
room cooling fan during the more protracted backup and restore sessions.

A five minute time out on a typical single drive windows PC is unlikely
to show any effect other than perhaps with older versions of windows such
as win2k and possibly winXP. From (the ironically named) Vista onwards,
system disk activity more or less guarantees that such a long time out on
spin down never gets a chance to kick in. :-)

Of course, when extra drives are used for data storage, these are more
likely to be left alone long enough for the spin down to kick in. Whilst
a couple of GB's of write cache can mask the effect on write accesses,
read accesses still remain at the mercy of such spin up delays so can
still be a source of frustration when accessing large media files for
playback or further processing.

Obviously, when it comes to such a trade off between 'instant
gratification' and energy savings, the choice is highly personal and,
therefore, beyond reasoned argument.

--
Johnny B Good