A word of warning about booting Windows from 2.5" drive

Silencing hard drives, optical drives and other storage devices

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Longwalker
Posts: 53
Joined: Sun Aug 06, 2006 2:35 pm

A word of warning about booting Windows from 2.5" drive

Post by Longwalker » Mon Jul 09, 2007 1:32 am

[Moderator edit: removed long post (in Latin?), presumably put there by the original poster. If the original poster wants to remove their posts and/or the the whole thread, then please PM one of the moderators.

Please see "Mr Evil's" post below, for what I think is a quote of the original post on this thread.]
Last edited by Longwalker on Wed Jul 11, 2007 9:55 pm, edited 1 time in total.

nick705
Posts: 1162
Joined: Tue Mar 23, 2004 3:26 pm
Location: UK

Post by nick705 » Mon Jul 09, 2007 1:58 am

hmmm... I can certainly see how it could be a problem, but when 2.5" drives are used for their primary purpose (ie in laptops), they'll invariably be the system drive and hence should be dying like flies... ?

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Jul 09, 2007 1:59 am

a couple of points:
1) there are several people on SPCR who have been using 2.5" as boot (I think) for several years (IsaacK, MikeC, Ed Ng off the top of my head), no failures reported.

2) increasingly people are now using laptops as their main machines, if this was true there should be hard drives dying all over the place

3) there are various apps to disable APM (see here), seems to work for Fujitsu and maybe other makes too.
when 2.5" drives are used for their primary purpose (ie in laptops), they'll invariably be the system drive and hence should be dying like flies... ?
the counter-arument is that laptops are not intended to be left on 24/7,so not so many head load/unload cycles.

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Post by Das_Saunamies » Mon Jul 09, 2007 3:30 am

Not to mention the power states a laptop goes into if left unused for just a few minutes. Desktops aren't usually configured to hibernate or cycle down very quickly, so they grind on for a good thirty minutes or more.

The disk access rates of Windows are infuriating regardless. Half the time I'm sitting there, doing nothing on my Satellite Pro the HDD light blinks and the drive taps in and out. It's like someone's trying to use it to emulate the sound a mechanical clock makes, tick in, tock out... tick in, tock out... ARGH! :lol:

I even uninstalled my antivirus software, but that just got rid of the massive seeks it performs, not the tick-tock-service-clock.

Even if these cycles happen(no solid data presented, although I am inclined to believe people's words), it seems the drives can be used successfully as boot drives regardless. It's good to discuss it though.

qviri
Posts: 2465
Joined: Tue May 24, 2005 8:22 pm
Location: Berlin
Contact:

Post by qviri » Mon Jul 09, 2007 4:53 am

jaganath wrote:1) there are several people on SPCR who have been using 2.5" as boot (I think) for several years (IsaacK, MikeC, Ed Ng off the top of my head), no failures reported.
My experience is not as extensive as some of the others' here, but I've been booting from a 2.5" Samsung MP0603H for two years. At least a year and a half of that was spent basically 24/7 (shutdowns for holidays and moving). No signs of impeding failure as of yet (cross fingers), SMART readings are fine.

I run Windows 2000 with a bbLean shell if that changes anything.

hmsrolst
*Lifetime Patron*
Posts: 427
Joined: Sun Aug 11, 2002 3:26 pm
Location: Arlington, VA USA

Post by hmsrolst » Mon Jul 09, 2007 5:46 am

I've been using Samsung MP0402H's as bootdrives on numerous desktop systems, some as long as 2+ years and on 24/7, so far with no problem.

Mr Evil
Posts: 566
Joined: Fri Jan 20, 2006 10:12 am
Location: UK
Contact:

Re: A word of warning about booting Windows from 2.5" d

Post by Mr Evil » Mon Jul 09, 2007 7:17 am

Longwalker wrote:...There's no practical way to keep Windows from touching the disk regularly without rendering the system unusable. There are just too many vital services to shut down...
I don't know about that. The server I put together for our LAN recently has its main drive spin down quite often (the power management is set to spin down drives after 30m of inactivity) The nLite'd version of XP that it runs is minimal, but usable for most things. It's certainly possible to cut down on disc access a lot without having any noticeable impact on functionality.

Longwalker
Posts: 53
Joined: Sun Aug 06, 2006 2:35 pm

Post by Longwalker » Mon Jul 09, 2007 10:24 am

[Moderator edit: removed long post (in Latin?), presumably put there by the original poster. If the original poster wants to remove their posts and/or the the whole thread, then please PM one of the moderators.]
Last edited by Longwalker on Wed Jul 11, 2007 9:56 pm, edited 1 time in total.

quikkie
Posts: 235
Joined: Tue Sep 20, 2005 5:21 am
Location: Soham, UK

Post by quikkie » Tue Jul 10, 2007 1:53 am

so what is it exactly about a head load/unload that is harmful to the disk?

I thought that with the "ramp" load/unload system that hitachi developed in the '90s that the wear you would see (compared with CSS type drives) was drastically reduced - a 2005 Hitachi white paper details that a travelstar drive managed in excess of one million load/unload ops. And it's not the head that travels up the ramp it's a tab that extends beyond the read/write head.



Note: I'm not so sad as to trawl the net for white papers to prove my point. I was looking for info on the power on behaviour of a disk i.e. does the head sweep the entire disk at power on or not as this will have a direct bearing on the recoverability of failed disk in my possession.

Longwalker
Posts: 53
Joined: Sun Aug 06, 2006 2:35 pm

Post by Longwalker » Tue Jul 10, 2007 9:55 am

quikkie wrote: details that a travelstar drive managed in excess of one million load/unload ops.
Assume Windows and the drive firmware conspire to force a load cycle every minute. If the drive is on 24/7, it'll use up all those million load cycles after a million minutes, or just under two years.

Now consider drives, like the Momentus 5400.3, that are only rated for 600,000 load cycles and that two load cycles per minute are quite possible. Work the numbers and the effective drive lifespan until exceeding cycle specification is only 5,000 hours, or 204 days 24/7. That's a problem.

Few drives will expire on the 600,001th load cycle, but running beyond specified life limits means dicing with an increased risk of sudden failure.

continuum
*Lifetime Patron*
Posts: 213
Joined: Tue Jul 19, 2005 10:23 pm
Location: California
Contact:

Post by continuum » Tue Jul 10, 2007 9:59 am

We haven't had any problems in production here, although FWIW, we do quite a bit of Linux here, and the Windows boxes here usually run extensively modified versions of Windows.

Hmm... I assume our customers have addressed this. Our failure rates indicate this isn't a problem, although I can't say how this has been addressed.

quikkie
Posts: 235
Joined: Tue Sep 20, 2005 5:21 am
Location: Soham, UK

Post by quikkie » Wed Jul 11, 2007 4:50 am

Longwalker wrote:Now consider drives, like the Momentus 5400.3, that are only rated for 600,000 load cycles
The travelstar quoted in the white paper was rated for 300K head load/unloads and managed 1million+. OTOH it may well have been the best drive of the batch.

I agree with you - people should look after their 2.5" disks and minimizing head loads sounds a reasonable way to help do that. Having said that I still don't know where the wear happens - I presume it's the pivot on the arm that the read/write head is mounted on that eventually wears out but I know know for certain.

IsaacKuo
Posts: 1705
Joined: Fri Jan 23, 2004 7:50 am
Location: Baton Rouge, Louisiana

Post by IsaacKuo » Wed Jul 11, 2007 6:15 am

Well I have had plenty of 2.5" drive failures. However, I got most of my 2.5" drives second hand, and given the abuse I put them through I really don't know what to attribute the failures to.

In any case, I mostly don't use Windows so the failures are certainly not due to anything Windows does.

Since I switched to diskless workstations, I've more or less stopped using 2.5" drives except for actual laptop computers. The performance of a local 2.5" hard drive is similar to the performance of diskless netbooting over a 100mbit LAN. The latter costs less, gives me more design flexibility, and consumes slightly less power. And it's easier to administer.

Aris
Posts: 2299
Joined: Mon Dec 15, 2003 10:29 am
Location: Bellevue, Nebraska
Contact:

Post by Aris » Wed Jul 11, 2007 6:16 am

i've been using a 2.5" drive for years, system runs 24/7, never had a problem. Ive actually have less problems with 2.5" drives than 3.5" drives since the larger drives are more prone to overheating.

Just cause you finally figured out how to read SMART data, and you noticed somthing strange doesnt mean its going to be catastrophic.

I know a lot of people that use notebooks as their main system, and use it pretty much all day. I've never heard of any of them having problems with their 2.5" disk drives, and all their laptops run on windows. Its not like theirs a special "notebook edition" widows installment or anything.

I've never seen a warning from any notebook manufacturer about not using a notebook continuously too long or risk harm to your hard drive, and they are pretty picky about making sure they cover their ass's about even the most mundane things that could possibly go wrong. Likewise ive never seen any such warning from disk drive manufacturers about using a drive "inproperly" by leaving your system running 24/7.

Theres lots of things that wear down the lifespan of electronics. People say you shouldnt put a sleeve fan in your PSU, cause it MIGHT only last 5 years instead of 10. You have to keep things in context, even IF you do shorten the lifespan of somthing, more than likely it will still last way longer than you will ever actually use it.

inti
Posts: 118
Joined: Thu Feb 10, 2005 7:09 am
Location: here

Post by inti » Wed Jul 11, 2007 6:34 am

I think the suggestion that Windows XP accesses the drive on average once every 30 seconds is misleading. My wife has a notebook running Windows XP SP2 with default configuration (she won't let me near it ...) and McAfee antivirus, and she often leaves it on, and idle obviously, overnight. In the quiet environment of my working room, I can hear it spin up the hard drive approximately once every 30 minutes.

Of course additional software, like instant messenger software or always-on email software, might change that. And if you are actively working on the PC, that is likely to require much more frequent drive access.

It is certainly possible to set up a Windows XP machine so that the boot drive spins down indefinitely when the machine is idle: it is just a question of disabling various non-essential Windows services and either disabling virtual memory or using a RAM drive for it (are more details needed here?). But if a typical application is running then it seems Windows needs to access the boot drive very regularly, maybe because the application loads and unloads DLLs continuously and Windows does not cache them efficiently. And sometimes just refreshing the desktop or opening the Start Menu needs the boot drive to be spun up, because Windows needs to access all the various application .EXE files to re-read the application icons, again because Windows does not cache icons efficiently.

vincentfox
Posts: 271
Joined: Sun Oct 09, 2005 8:35 pm
Location: CA

Post by vincentfox » Wed Jul 11, 2007 8:04 am

Dumb question:

Can we disable AAM or power-saving features to mitigate this risk?

WR304
Posts: 412
Joined: Tue Sep 20, 2005 1:21 pm
Location: UK

Post by WR304 » Wed Jul 11, 2007 8:42 am

inti wrote:In the quiet environment of my working room, I can hear it spin up the hard drive approximately once every 30 minutes.
One easy way to reduce hard disk accessing in Windows XP is to disable the auto defrag feature.

You can simply download the Microsoft TweakUI application and deselect the "Optimize hard disk when idle" option:

http://www.microsoft.com/windowsxp/down ... rtoys.mspx

It's under the general tab and goes a long way to stopping the drive making a racket when the PC is idle. :)

Wait until your wife isn't looking... :wink:

Image
Last edited by WR304 on Wed Jul 11, 2007 11:15 am, edited 2 times in total.

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Post by Das_Saunamies » Wed Jul 11, 2007 10:14 am

I've got that disabled, but my Satellite Pro tick-tocks regardless. Probably because it's an old model with only 384MB of RAM. When it dozes off the sound stops, and thankfully that's quite often.

Longwalker
Posts: 53
Joined: Sun Aug 06, 2006 2:35 pm

Post by Longwalker » Wed Jul 11, 2007 7:23 pm

If a particular configuration does have a regular disk access problem, though, then you are at risk of wearing out a 2.5" disk in short order. It's no skin off my nose if anyone is ignorant or arrogant enough (like Aris here...) to wear down their drive--you're just costing yourself money and risking your data but you're not costing me anything....

peteamer
*Lifetime Patron*
Posts: 1740
Joined: Sun Dec 21, 2003 11:24 am
Location: 'Sunny' Cornwall U.K.

Post by peteamer » Wed Jul 11, 2007 11:13 pm

My Sony Vaio laptop is somewhere around four years old. It folds from 6PM to ~8AM during the week (Mon-Thur) and non stop Fri evening-Mon morning with 1/2HR writing of progress to the HD for resume points.

Still on the original 20GB HD (Toshiba IIRC).

Just to throw some 'data' into the mix.....

Aris
Posts: 2299
Joined: Mon Dec 15, 2003 10:29 am
Location: Bellevue, Nebraska
Contact:

Post by Aris » Wed Jul 11, 2007 11:47 pm

Longwalker wrote:If a particular configuration does have a regular disk access problem, though, then you are at risk of wearing out a 2.5" disk in short order. It's no skin off my nose if anyone is ignorant or arrogant enough (like Aris here...) to wear down their drive--you're just costing yourself money and risking your data but you're not costing me anything....
still waiting to see some proof.

a forum full of angry notebook drive customers?
a notebook manufacturer with warnings or warrenty fine print stating somthing about extended use of system may damage drive?
a disk drive manufacturer with warnings or warrenty fine print stating somthing about extended use of system may damage drive?

all i see is a forum full of happy silent pc users with no issue's after years of operation, and a single individual with a datasheet but no evidence to back up his claims. you can speculate till your blue in the face about how certain drive activities may or may not effect the drives lifespan, but it means nothing if you have no way to prove it. why dont you email seagate or samsung about your "discovery" and ask them what they think about it.

Phlebas
Posts: 19
Joined: Tue Jan 17, 2006 8:08 am

Post by Phlebas » Thu Jul 12, 2007 12:15 am

I've got a couple of Western Digital WD1600BEVS drives in my PC. The one with the Windows partition on it is averaging around 155 load cycles per day. The drives are rated for 600,000 cycles so I guess I'll need to start getting concerned in 2017... 8)

FWIW my non-windows drive averages around 30 load cycles per day.

nick705
Posts: 1162
Joined: Tue Mar 23, 2004 3:26 pm
Location: UK

Post by nick705 » Thu Jul 12, 2007 1:32 am

Aris wrote: why dont you email seagate or samsung about your "discovery" and ask them what they think about it.
Whatever the rights and wrongs of the argument, this thread was a perfectly reasonable and civilised discussion until you jumped in with:
Aris wrote:Just cause you finally figured out how to read SMART data, and you noticed somthing strange doesnt mean its going to be catastrophic. Its like all those noobs that come on this forum that just figured out how to check their CPU temps and freak out when its 70c and think their CPU is going to fry.
There was some more stuff, including ordering the OP not to "come on here crying wolf" or words to that effect, but you've since edited it out.

I see the OP has now deleted his original posts, and frankly I don't blame him. I for one am absolutely sick to the back teeth with rude, obnoxious know-it-alls who think they can be as arrogant and offensive as they like on a web forum, safe in the knowledge they won't get the richly deserved punch in the mouth their behaviour would earn them in real life. There's more than enough of this at HardForum and other hangouts for spotty adolescents with no social skills, god knows we can do without it here.

It's unnecessary, uncalled-for and just f#$%ing bad manners.

Aris
Posts: 2299
Joined: Mon Dec 15, 2003 10:29 am
Location: Bellevue, Nebraska
Contact:

Post by Aris » Thu Jul 12, 2007 1:55 am

what i find offensive is when people post on the forum about personal opinions and try to pass it off as factual. i'm all for a good debate and open discussion about discoveries, but if thats what it is then post it in that manner like "hey i found somthing interesting blah blah, what do you think?" not "warning about booting with 2.5" drives".

Is his information for possible drive falure plausible? Sure
Would i like to see further investigation into this? Absolutely

Hell i'd even like to see some sort of review, or reply statement from a drive manufacturer about this. But he didnt do any of those things. He noticed somthing odd, and jumped to the conclusion that it was somehow catastrophic in nature to use these drives for boot devices, and posted on a forum stating it as factual. This is exactly how rumors get started, and its bad form.

and i didnt edit my post, i think the moderator did, and i think the moderator edit'd his posts also because he had said somthing fairly vulgur as a response to my post.

highlandsun
Posts: 139
Joined: Thu Nov 10, 2005 2:04 am
Location: Los Angeles, CA
Contact:

Post by highlandsun » Thu Jul 12, 2007 3:09 am

Once every 30 minutes, I wish. Try this - download and install TaskInfo on your Windows machine:

http://www.iarsn.com/taskinfo.html

One of the status displays it shows is how many pageins, pageouts, reads, and writes per second are occurring. You'll notice that even on a completely idle machine with no apps loaded (not even background apps) the I/O counters never stay at zero. Another handy thing about this program is that it can show you all the open files in the system, as well as their read/write pointer locations. If you track these down, you'll see that a number of Windows kernel services (like the LSA) are continuously writing to their respective log files, and there is no way to disable that logging activity. It's only a couple bytes per second, but eventually it forces a disk flush. Another thing you'll see, if you have the Task Scheduler running, is that your disks spin up every 5 minutes if you have WindowsUpdate enabled. When I still ran Windows regularly, I turned off as many of these things as possible, but still my laptop's drive would spin up once every 5-10 minutes. Windows totally sucks.

Now that I run Linux, and again I've disabled all the services I don't care about, my idle machine is actually idle when I walk away from it, and the drive actually gets to spin down and stay down for hours at a time. (Note: my laptop has 2GB of RAM installed; almost all of my work gets cached and stays in RAM thereafter. On Windows XP the braindead memory manager flushes the cache to disk all the time, even though there's 1.5GB of free RAM in the system. It actually checks every 1/4 second for any pages it can flush, thus defeating all benefit of having more than 512MB of RAM in the machine.)

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Thu Jul 12, 2007 3:39 am

vincentfox wrote:Dumb question:

Can we disable AAM or power-saving features to mitigate this risk?
not dumb at all. :wink: yes, on some drives. see the link in my post, third from the top.

Mr Evil
Posts: 566
Joined: Fri Jan 20, 2006 10:12 am
Location: UK
Contact:

Post by Mr Evil » Thu Jul 12, 2007 5:28 am

So that the discussion doesn't become completely confusing and useless for future readers just because someone can't handle criticism, the original post was:
Longwalker wrote:Something that seems to be missing from the regular talk on SPCR about 2.5" drives is their long-term life when used as boot drives. There is a real problem in this area because of the drives' head unloading 'feature' and Windows' boot disk access habits. The net result is that 2.5" drives used for boot won't last very long.

One part of the problem lies in drive firmware. In the interests of saving power in laptops, many 2.5" drives are designed to unload (park) their heads after some seconds of inactivity. For example, Momentus 5400.x drives unload after about 10-15 seconds of inactivity. This feature is likely to be found in 7200.x series drives and those from other manufacturers as well. There is no known way to disable this feature on Seagate drives.

On its own, rapid head unloading isn't much of a problem. It only becomes a problem when combined with a misfeature in Windows.

The Windows kernel and core services effectively conspire to access the system (boot) drive several times a minute. On a normal Windows install, it's rare for the OS to leave the system drive alone for more than 30 seconds. Windows Management Instrumentation is a big culprit here, but both the kernel and many other services regularly access the system drive. There's no practical way to keep Windows from touching the disk regularly without rendering the system unusable. There are just too many vital services to shut down.

Put together, rapid head unloading and Windows' habitual disk access mean that a 2.5" system drive will be forever cycling its heads from the parked to loaded position. The drive firmware will park the drive heads after 15 seconds of inactivity. After perhaps 20 seconds, however, Windows will want access to the disk. Wash, rinse, repeat, to the worst case tune of multiple head load cycles per minute. This level of cycling will kill a drive in very little time at all. Indeed, on a 2.5" boot drive left on 24/7, even casual observation of SMART data will show the drive literally cycling itself to death day-by-day.

The bottom line is that 2.5" disks should not be used for booting Windows unless a short lifespan is acceptable and no other options are available.
This has been an interesting and informative thread and I want it to stay that way.

inti
Posts: 118
Joined: Thu Feb 10, 2005 7:09 am
Location: here

Post by inti » Thu Jul 12, 2007 6:19 am

highlandsun thank you for your informative post.


The PLA service you referred to is Performance Logs and Alerts Service? One of the many I have disabled, on my own machine :)

I have also disabled Indexing Service. And for a nice idle machine, I think you need to either disable 'Warn when low on disk space' in the System Restore settings, or disable the System Restore Service completely.

Another tip for reducing drive accesses is to set Windows NTFS not to update the File Accessed Time on each file that is accessed. I think if you don't do that, then whenever Windows reads from a file (including a file that is cached in RAM) then it has to write to the drive to update the File Accessed Time - and most users do not have Write Caching enabled.

I found a very detailed and technical Microsoft document all about disk performance and caching - it has a lot of information which I've not seen elsewhere:
http://download.microsoft.com/download/ ... s_perf.doc

---

On my Windows XP machine with 2GB RAM, I have tried totally disabling virtual memory, and also tried putting a small amount of virtual memory on a RAM drive.

Total disablement was mostly OK, but eventually Windows would create some virtual memory for itself on the C: drive - it seems there are some situations in which Windows absolutely needs to have some virtual memory, sorry I cannot be more specific. Once it has created the virtual memory on the C: drive then you are stuck with that unless you disable it again. (I have also read articles saying that Windows should never be run with zero virtual memory.)

Putting the virtual memory on a RAM drive has worked great for me for 18 months now. I have given it a fixed pagefile size of 64MB and it has never seemed to need more. I recommend Ramdisk Plus from SuperSpeed for this, since as far as I know it is the only Windows RAM drive which allows you to place the pagefile in RAM so that there is no pagefile left on any physical hard drive. (Other RAM drive products create the RAM drive too late in the boot process so that Windows will create a pagefile on the C: drive during bootup, thinking it has no other pagefile.)

NeilBlanchard
Moderator
Posts: 7681
Joined: Mon Dec 09, 2002 7:11 pm
Location: Maynard, MA, Eaarth
Contact:

Post by NeilBlanchard » Thu Jul 12, 2007 7:17 am

Hello SPCR posters,

The concern for the continued civility of persons on this thread has been raised, and I will be watching it closely. Please beware of letting your ego get in the way of the discussion -- and keep it civil!

[/moderator's hat]

Question to the original poster: how do you explain the fact that (virtually) every laptop in existence uses a 2.5" hard drive as it's boot drive, apparently without any undue stress?

Bluefront
*Lifetime Patron*
Posts: 5316
Joined: Sat Jan 18, 2003 2:19 pm
Location: St Louis (county) Missouri USA

Post by Bluefront » Thu Jul 12, 2007 7:43 am

Neil's last question brings us back to the often-asked "do laptop drives last as long as the bigger drives?". My take.....In the past, laptop drives do not last as long as 3.5" drives for a number of reasons, mostly involving the physical characteristics of laptop usage involving hard drives. Here I'm talking about heat, vibrations, movements......all of which affect the lifespan of a laptop drive.

Now take a new modern laptop drive.....put it in a normal computer with cooling, relatively stable mounting, no movements while the drive is running. Will this drive last as long as a 3.5" drive under the same conditions? Tough question.....no answer from me. I'm using several laptop drives like this, successfully so far. :)

Post Reply