AVS › AVS Forum › Video Components › Home Theater Computers › Spin hard drives down or leave them running?
New Posts  All Forums:Forum Nav:

Spin hard drives down or leave them running?

post #1 of 52
7/19/12 at 7:10am
Thread Starter 
Sorry if this has been asked before, but many of the posts on AVS looked really old.

I noticed that when I try to play a video that's stored on my WHS 2011 media server on an XBMC client, I usually have to wait about 5-10 seconds for it to start playing. After that, they play almost immediately. I'm guess this is because the hard drives spin down when they're not in use for while.

So I'm wondering, do you guys usually let your hard drives spin down, or is it okay to leave them running all the time? Would running them all the time decrease longevity, and would the increased power demands make a big difference on the electricity bill (for say, 6 hard drives)?
Reply
Reply

AVS Top Picks

post #2 of 52
7/19/12 at 7:16am
I have the same issue when the server has been sitting around for a while unused.

it's on... but in energy save mode.

FLEXRAID I believe lets you spin them down.

I don't mind waiting for the first one a couple seconds. Figure it's worth the energy savings.
Reply
Reply
post #3 of 52
7/19/12 at 8:03am
Thread Starter 
Well I did find out about an extensive hard drive study google did in 2007. They essentially found that hard drive life had more to do with with manufacturer, model number, and age than usage patterns. However, the paper says these hard drives are left spinning constantly, so it obviously doesn't differentiate between spinning down or leaving on all the time. I'm guessing google also uses server grade drives and not 'green' drives like I use.

The other question is how much it costs to keep them spinning. I've been googling it for a while, but I haven't found a great answer. The amount of electricity a hard drive uses doesn't seem to be trivial though, and they use far less electricity when they're asleep. It might be best then to just let them sleep.
Reply
Reply
post #4 of 52
7/19/12 at 8:09am
Quote:
Originally Posted by lockdown571 View Post

Well I did find out about an extensive hard drive study google did in 2007. They essentially found that hard drive life had more to do with with manufacturer, model number, and age than usage patterns. However, the paper says these hard drives are left spinning constantly, so it obviously doesn't differentiate between spinning down or leaving on all the time. I'm guessing google also uses server grade drives and not 'green' drives like I use.
The other question is how much it costs to keep them spinning. I've been googling it for a while, but I haven't found a great answer. The amount of electricity a hard drive uses doesn't seem to be trivial though, and they use far less electricity when they're asleep. It might be best then to just let them sleep.

Here are the power usage results of the 8 drive Goliath Server that I built for Flocko...

http://forum.xbmc.org/showthread.php?tid=131865&pid=1142693#pid1142693

http://www.assassinserver.com/?p=130
Reply
Reply
post #5 of 52
7/19/12 at 8:10am
Yes, I let my HDD spin down.
Reply
Reply
post #6 of 52
7/19/12 at 8:14am
Quote:
Originally Posted by dksc318 View Post

Yes, I let my HDD spin down.

+1.
Reply
Reply
post #7 of 52
7/19/12 at 8:44am
Thread Starter 
Quote:
Originally Posted by assassin View Post

Here are the power usage results of the 8 drive Goliath Server that I built for Flocko...
http://forum.xbmc.org/showthread.php?tid=131865&pid=1142693#pid1142693
http://www.assassinserver.com/?p=130

Thanks assassin. Can you give some more details though? Were the hard drives spun down at idle? Also, when playing a movie, was only one hard spinning up, or were all of them?
Reply
Reply
post #8 of 52
7/19/12 at 8:51am
Thread Starter 
With a bit more searching, it sounds like a 3.5" hard drive typically uses 10-15 watts while spinning. Using this calculator, I found that 4 hard drives spinning constantly at 15 watts each would use about $4.75/month. That's not too bad.
Reply
Reply
post #9 of 52
7/19/12 at 10:06am
UnRAID lets the drives spin down when not in use. When I need to access a file it only spins up the drive the file is located on. If I add files to a drive it also spins up the parity drive so it can make changes to the parity disc to keep it current. Sleeping drives not only consume less energy but they produce less heat. Less heat equates to better life expectancy. I keep the server up and running 24/7 for months on end and have only experienced two drive failures over a five year period (one was the parity drive, which obviously gets exercised more than the rest of the array).
Reply
Reply
post #10 of 52
7/19/12 at 10:11am
I let mine spin down also. Xbmc has a adding that will wake LAN . You plug in the servers Mac address and that's it. This will pretty much spin up my hard drives while Xbmc launches.
Reply
Reply
post #11 of 52
7/19/12 at 10:12am
Quote:
Originally Posted by lockdown571 View Post

Thanks assassin. Can you give some more details though? Were the hard drives spun down at idle? Also, when playing a movie, was only one hard spinning up, or were all of them?

No problem. FlexRaid and drives spin down when not in use.
Reply
Reply
post #12 of 52
7/19/12 at 10:19am
Quote:
Originally Posted by lockdown571 View Post

With a bit more searching, it sounds like a 3.5" hard drive typically uses 10-15 watts while spinning. Using this calculator, I found that 4 hard drives spinning constantly at 15 watts each would use about $4.75/month. That's not too bad.

But it is still like $50 a year ..
Reply
Reply
post #13 of 52
7/19/12 at 10:57am
Thread Starter 
Quote:
Originally Posted by Mfusick View Post

But it is still like $50 a year ..

Certainly. Honestly, it doesn't really bother me that much (waiting for the hard drive to spin up), but I notice it tends to frustrate other users. XBMC is so responsive, so I think people find it jarring when they press play and nothing happens for 5 seconds. I think it might be helpful if there were some sort of blinking "your movie is loading" screen, and maybe some of the XBMC skins implement this.
Reply
Reply
post #14 of 52
7/19/12 at 11:17am
Quote:
Originally Posted by captain_video View Post

UnRAID lets the drives spin down when not in use. When I need to access a file it only spins up the drive the file is located on. If I add files to a drive it also spins up the parity drive so it can make changes to the parity disc to keep it current. Sleeping drives not only consume less energy but they produce less heat. Less heat equates to better life expectancy. I keep the server up and running 24/7 for months on end and have only experienced two drive failures over a five year period (one was the parity drive, which obviously gets exercised more than the rest of the array).
ah, but the spinning up/powering up after being sleepy causes more wear on any electronic device... even our beloved HDD's...

but in the end I would call the wear issue a wash... I think... or close enough to even to not be an issue...

but I let my unraid drives spin down too... any saved energy $$ is a good thing...
Reply
Reply
post #15 of 52
7/20/12 at 6:46am
Average drives maybe 5-12W idle, but average drives are not appropriate for media stores/HTPCs.
WD Green 2TB takes 3.7 W spinning idle ( http://hothardware.com/Reviews/Western-Digital-2TB-Caviar-Green-Power-Hard-Drive/ )
I am hoping that the WD Red drives take even less, since they are designed for NASes.
Reply
Reply
post #16 of 52
7/20/12 at 7:01am
Quote:
Originally Posted by Somewhatlost View Post

ah, but the spinning up/powering up after being sleepy causes more wear on any electronic device... even our beloved HDD's.

That's the real question, IMO. But is this statement true? Does one spin-up/spin-down produce more wear than, say, 12 hours of continuous spin? I have my doubts, but certainly don't know.
Reply
Reply
post #17 of 52
7/20/12 at 7:50am
Quote:
Originally Posted by Somewhatlost View Post

ah, but the spinning up/powering up after being sleepy causes more wear on any electronic device... even our beloved HDD's...
but in the end I would call the wear issue a wash... I think... or close enough to even to not be an issue...
but I let my unraid drives spin down too... any saved energy $$ is a good thing...

Probably but not enough to worry about.
Reply
Reply
post #18 of 52
7/20/12 at 8:00am
Quote:
Originally Posted by lockdown571 View Post

With a bit more searching, it sounds like a 3.5" hard drive typically uses 10-15 watts while spinning. Using this calculator, I found that 4 hard drives spinning constantly at 15 watts each would use about $4.75/month. That's not too bad.

Buy a Kill-A-Watt power meter and you can calculate how much energy each drive will use.
Run your PC with no drives connected and measure the power. Then add in one drive at a time and take measurements for each drive.
When all the drives are idle, measure the power and calculate the difference between full load and idle.
Reply
Reply
post #19 of 52
7/20/12 at 8:33am
Quote:
Originally Posted by DanPackMan View Post

That's the real question, IMO. But is this statement true? Does one spin-up/spin-down produce more wear than, say, 12 hours of continuous spin? I have my doubts, but certainly don't know.

I doubt its true. Just like everything (it seems) in HTPC I doubt its been tested.
Reply
Reply
post #20 of 52
7/20/12 at 9:48am
Quote:
Originally Posted by assassin View Post

I doubt its true. Just like everything (it seems) in HTPC I doubt its been tested.
Agree. The only thing we can say with any certainty is that keeping your drives spinning will consume more energy. That's why I spin my drives down when not in use because there's no doubt on whether or not I'm saving money. It could hurt or help drive life but who knows, so why avoid something that may not even be a problem?
Reply
Reply
post #21 of 52
7/20/12 at 10:01am
I have them spin down after 30 min since I have a lot of HDDs. They use a surprising amount of electricity, when buying a power supply how many HDDs you have is one of the main factors in how big of one you need.

I also strategically place data on them to further lower power consumption, on the same drive I will make 2 partitions, one for Windows and one for WMC recording. If I can fit anything else that is commonly used on that drive then I will. Then I will put less used stuff like games on the other drives so they can be off 90% of the time.
Reply
Reply
post #22 of 52
7/20/12 at 1:23pm
Quote:
Originally Posted by Mfusick View Post

Probably but not enough to worry about.
I tend to agree...
especially in my usage scenario, where the drives are all idle for most of the week, and I only watch a movie or two on the weekends... makes far more sense to let the drives sleep...
Reply
Reply
post #23 of 52
7/20/12 at 2:02pm
Quote:
Originally Posted by Somewhatlost View Post

I tend to agree...
especially in my usage scenario, where the drives are all idle for most of the week, and I only watch a movie or two on the weekends... makes far more sense to let the drives sleep...

Which is primarily how Goliath is used .

I really just can't see a difference in the power bill . We are in the really dog *ss hot part of the year here and the ac is running non stop but I still don't see much if any extra usage from the server on the power bill .
Reply
Reply
post #24 of 52
7/21/12 at 5:21am
Quote:
Originally Posted by hirent View Post

Buy a Kill-A-Watt power meter and you can calculate how much energy...
A nicer product is Belkin's Conserve Insight--the display is remote from the plug-in.

These devices probably won't help all that much to answer the topic question, but it may be interesting info to have. At least, if you run all drives 24/7 then the wattage/cost will be accurate. Another thing to keep in mind is that watts translate directly to heat i.e. there is a cooling penalty to be paid.

I was surprised recently to learn that a TiVo with wireless AP and Belkin UPS draws only 70 watts. That's Good. But of course it's almost exactly like having a 75watt incandescent light bulb on 24/7 in that room also, which is Not so good.

My opinion is that frequent starts/stops of spinning hard drives, such that they may be started-up again when they are warm (or even hot) is not good for their motors. But if your server sits idle most of the day and is used for some small fraction of the 24hours, then you oughta spin them down, or turn them off entirely.
Reply
Reply
post #25 of 52
7/21/12 at 7:16am
Quote:
Originally Posted by assassin View Post

I doubt its true. Just like everything (it seems) in HTPC I doubt its been tested.
I am sure you are right.... HDD's, unlike ANY other bit of electronics or electromechanical gear ever made are created out of some special alien tech and black magic and are therefore not bound by the same laws of physics as regular electronics...

odd that S.M.A.R.T. keeps track of Start/Stop Count (power & sleep cycles all combined into one) cycles though... I am sure they do that just out of curiosity though, as it is not a 'Critical' count like a Read Error or anything...

so just out of curiosity, since HDD's can't be using any capacitors, resistors, diodes, inductors, motors, etc as they are all susceptible to the damaging effects of inrush current, rapid temp changes,etc(ie all the crap that happens during your average power on cycle) over time, just what are HDD's actually made of?

we obviously need to study what is a well known fact of electronics and physics, and is one the basic forms of HALT testing, just because we added 'HTPC' to it... I am sure 'HTPC' changes everything...
Edited by Somewhatlost - 7/21/12 at 8:46am
Reply
Reply
post #26 of 52
7/21/12 at 8:04am
Quote:
Originally Posted by Somewhatlost View Post

I am sure you are right.... HDD's, unlike ANY other bit of electronics or electromechanical gear ever made are created out of some special alien tech and black magic and are therefore not bound by the same laws of physics as regular electronics...
odd that S.M.A.R.T. keeps track of Start/Stop Count (power & sleep cycles all combined into one) cycles though... I am sure they do that just out of curiosity though, as it is not a 'Critical' count like a Read Error or anything...
so just out of curiosity, since HDD's can't be using any capacitors, resistors, diodes, inductors, motors, etc as they are all susceptible to the damaging effects of inrush current, rapid temp changes,etc(ie all the crap that happens during your average power on cycle) over time, just what are HDD's actually made of?

I think you should a 60 months test and report back every 12 months to us to let us know. Use all the major manufacturers and options. Also run each on a server, HTPC and PC to make sure you get a good variety of data to comb through.
Reply
Reply
post #27 of 52
7/21/12 at 8:05am
Oh, and I am pretty sure that they are made of dragon scales. Seems as likely as some of the other preposterous things reported.
Reply
Reply
post #28 of 52
7/21/12 at 8:15am
Quote:
Originally Posted by assassin View Post

I doubt its true. Just like everything (it seems) in HTPC I doubt its been tested.

The main reason that reliability regarding spin-down hasn't been tested is because the sample sizes just aren't big enough. The folks who have enough drives to provide useful data largely aren't going to be spinning down drives anyway, because they need the throughput of the drives 24/7 (big datacenters and such). The only people who would have access to enough data to provide a clear cut answer that isn't based on speculation are the manufacturers, who record plenty of data about every failed drive they get in RMA. And believe me, they would never, ever admit to an issue like that unless it was overwhelmingly obvious that spun-down drives were failing much earlier than drives kept spinning.

If you are spinning up and spinning down your drives 20 times a day, 365 days a year, you are probably going to encounter problems due to the constant heating and cooling going on. But who actually has a usage pattern like that? Most likely, spin-up/spin-down is limited to 2-3 times a day, if that. Thermal expansion of electronic components is widely documented as a root cause of failure over time, but I really can't see a usage pattern for hard drives that would produce a noticeable difference in drive life unless you're intentionally abusing it. Perhaps it would reduce the life of a drive from six or seven years down to five or six, but most of us are replacing drives before then anyway. If WD's predictions are accurate, we'll all be rocking 20TB+ drives by 2016 anyway.
Reply
Reply
post #29 of 52
7/21/12 at 8:36am
Quote:
Originally Posted by mcturkey View Post

The main reason that reliability regarding spin-down hasn't been tested is because the sample sizes just aren't big enough. The folks who have enough drives to provide useful data largely aren't going to be spinning down drives anyway, because they need the throughput of the drives 24/7 (big datacenters and such). The only people who would have access to enough data to provide a clear cut answer that isn't based on speculation are the manufacturers, who record plenty of data about every failed drive they get in RMA. And believe me, they would never, ever admit to an issue like that unless it was overwhelmingly obvious that spun-down drives were failing much earlier than drives kept spinning.
If you are spinning up and spinning down your drives 20 times a day, 365 days a year, you are probably going to encounter problems due to the constant heating and cooling going on. But who actually has a usage pattern like that? Most likely, spin-up/spin-down is limited to 2-3 times a day, if that. Thermal expansion of electronic components is widely documented as a root cause of failure over time, but I really can't see a usage pattern for hard drives that would produce a noticeable difference in drive life unless you're intentionally abusing it. Perhaps it would reduce the life of a drive from six or seven years down to five or six, but most of us are replacing drives before then anyway. If WD's predictions are accurate, we'll all be rocking 20TB+ drives by 2016 anyway.

Those are my thoughts as well.
Reply
Reply
post #30 of 52
7/21/12 at 9:20am
Quote:
Originally Posted by assassin View Post

Those are my thoughts as well.
I am confused...
before, it appeared you were claiming that HDD's were somehow special magical beasties that did not suffer any wear from power/sleep cycling... which is obviously either insane or delusional...
but now, it appears that you admit they do wear when waking up/powering up, it is just not enough wear in a standard usage scenario to actually have any measurable effect on the drives life... which is accurate to the best of my knowledge...

I would probably go so far as to say that letting a drive sleep while not in use for extended periods of time (hours maybe, days definitely) probably might even help extend a drive's useful life in normal (well what the average HTPC/home user considers normal at least) usage, but really in something like 99.99999% of all cases, the drive will get replaced due to size or speed upgrades long before it fails regardless of sleepiness... (again in an average HTPC/home user environment... a datacenter could be a completely different scenario)

so just for clarity, what is assassin's official position?
A) HDD's are magical beasties that don't wear do to power/wake up?
or
B) HDD's are just like any other electromechanical device, and they do wear due to power/wake up, but that the wear from power/wake up is <= to the benefit of not wearing while sleeping (and the power savings are a huge bonus too)?
or
C) ???
Reply
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Home Theater Computers

AVS Top Picks

AVS › AVS Forum › Video Components › Home Theater Computers › Spin hard drives down or leave them running?