Service life for hard drives

Discussion in 'Technical Corner' started by Pi1otguy, May 3, 2012.

  1. Pi1otguy

    Pi1otguy Pattern Altitude

    Joined:
    Oct 24, 2007
    Messages:
    2,104
    Location:
    Long Beach, CA

    Display name:
    Fox McCloud
    I'm working with a few dozen 1 yr old desktops and the secondary hard drives are starting to fail. All the drives are the same brand, bought & installed at about the same time and started failing after 11 months of use But at the same time, the primary drives (different brand & size) have had no issues.

    What is a normal service life for new hard drives?
     
  2. denverpilot

    denverpilot Taxi to Parking

    Joined:
    Nov 8, 2009
    Messages:
    32,603
    Location:
    Denver, CO

    Display name:
    DenverPilot
    Manufacturer's website will have warranty information. Most are 3 years these days, with some "server quality" drives getting a longer warranty for a significantly (waste of money) higher price, and lots have been down as low as 1 year in the pricing battles a few years ago.
     
  3. denverpilot

    denverpilot Taxi to Parking

    Joined:
    Nov 8, 2009
    Messages:
    32,603
    Location:
    Denver, CO

    Display name:
    DenverPilot
    Oh... that was warranty information -- service life is even more all over the place. Most folks I know who track it are tracking specific LOTS of drives they buy in bulk. They typically fail in similar timeframes, even same drive, same everything... except different timeframe of purchase.

    At least one government agency near me here is replacing all disks on a rotating three year schedule, which to me, is excessive.

    We have everything from 10 year old (yeah, really... easy in fact) drives in Production to brand new ones fail.

    The U-shaped curve of "infant mortality" where more new drives fail than middle-aged ones, and then old ones start failing again on the upper right side of the "U", always applies.

    You mentioned these were secondary drives... added to the machines? Proper airflow/cooling? Heat kills drives more than anything. I have stuff spinning that's well past 10 years old... they're in temperature controlled environments, and they're rarely shut down. As long as the bearings hold out, they'll be there.
     
  4. John221us

    John221us En-Route

    Joined:
    Jan 5, 2012
    Messages:
    3,313
    Location:
    Rocklin, CA

    Display name:
    John
    +1, heat will significantly shorten drive life. Other than that, you can get bad batches, but expect at least 3 to 5 years of constant use (like a server). You used to get head drift and have to reformat, but newer technology compensates for that and you can't manually reformat anymore.
     
  5. Pi1otguy

    Pi1otguy Pattern Altitude

    Joined:
    Oct 24, 2007
    Messages:
    2,104
    Location:
    Long Beach, CA

    Display name:
    Fox McCloud
    Luckily we are under warranty. It's just that we're a failure shy of a 10% failure rate in only a year and the logistics of quickly swapping hardware when some desktops are 200+ miles away get's weird.
     
  6. denverpilot

    denverpilot Taxi to Parking

    Joined:
    Nov 8, 2009
    Messages:
    32,603
    Location:
    Denver, CO

    Display name:
    DenverPilot
    Not just weird, but expensive too... unless the manufacturer is paying all the shipping. Sometimes new drives are cheaper at numbers like 10%.
     
  7. Pi1otguy

    Pi1otguy Pattern Altitude

    Joined:
    Oct 24, 2007
    Messages:
    2,104
    Location:
    Long Beach, CA

    Display name:
    Fox McCloud
    Yes. The vendor added these hard drives. Typically, when I check the temps during high usage times we're seeing HD temps between 36C and 40C . Obviously, cooler would be better but that should be well within acceptable limits.
     
  8. jesse

    jesse Administrator Management Council Member

    Joined:
    Oct 2, 2005
    Messages:
    15,304
    Location:
    Lincoln, NE

    Display name:
    Jesse
    Typical consumer hard drives will have a MTBF from 500,000 hours up to a million or so. I wouldn't put too much faith in those numbers though.

    Drive failure rates are typically measured in the single digit percent range when many drives are analyzed. I can say that is about accurate for my company.

    For more information then you'd really like to know:
    http://static.usenix.org/events/fast07/tech/schroeder/schroeder_html/index.html
     
    Last edited: May 3, 2012
  9. Palmpilot

    Palmpilot Touchdown! Greaser!

    Joined:
    Apr 1, 2007
    Messages:
    11,366
    Location:
    PUDBY

    Display name:
    Richard Palm
    For the past few years I've been using Western Digital Caviar Green drives, because they run cooler, and I figure that lower temps will correlate with longevity. So far, so good.
     
  10. CT4ME

    CT4ME Cleared for Takeoff

    Joined:
    Oct 24, 2011
    Messages:
    1,138
    Location:
    Phoenix, AZ

    Display name:
    CT4ME
    There are certain brand/model/lots that have higher failure rates. Every wonder how Tiger or Newegg can have hundreds of certain drives for sale as "refurbs"? 'Could also be outside conditions, like bad power or static (if they were sensitive to that), vibrations, or even the power-saving BS. I can't help but think a hard drive turning itself on/off throughout the day isn't good.
     
    Last edited: May 4, 2012
  11. MN_Flyer

    MN_Flyer Pre-Flight

    Joined:
    Jul 8, 2011
    Messages:
    85
    Location:
    Saint Paul, MN

    Display name:
    MN_Flyer
    I monitor the health of my hard drives at work with this program:
    http://www.hdsentinel.com/

    There is a free trial version for private use that has no expiration date and just has a reminder to register (pay) at some time. There is a small fee for businesses or for a full-featured version.

    It runs in the system tray and shows you the status of all your hard drives. It monitors the SMART feedback from all your drives and alerts the user when there are errors on a disk. You can use this to see if your hard drives are operating at a safe temperature and it also monitors the real-time transfer of data onto and off of the drive and detects any errors.

    I have had very good luck with it. As long as your drives are not experiencing some type of sudden, physical failure, this utility should alert you that the drive health is declining and allow you enough time to copy the data onto a new hard drive.
     
  12. Skip Miller

    Skip Miller En-Route

    Joined:
    Feb 22, 2005
    Messages:
    3,977
    Location:
    New York City

    Display name:
    Skip Miller
    What is the failure mode for most hard drives? Is it the motor/disk bearings or the read/write head mechanism, or software? -Skip
     
  13. denverpilot

    denverpilot Taxi to Parking

    Joined:
    Nov 8, 2009
    Messages:
    32,603
    Location:
    Denver, CO

    Display name:
    DenverPilot
    Yes, yes, not often. ;)

    Specifically, bearings usually on the first. Motors seem to always work. Rare to hear motors making odd noises.

    Stuff getting out of spec and not moving to precise locations on the second. Or high error rates caused by same, actually. (The drive firmware tracks this.)

    Once in a while a component fails (usually due to excessive heat) in the electronics. I've seen a few drives brought back to life in dire situations by finding an identical drive and swapping the electronics.

    Unfortunately that trick often can't work with modern drives that track physical errors and won't have the physical "map" of the drive if you swap the electronics. A recovery lab can often figure that mess out for a price, though.

    The other failure mode I have seen was a manufacturing defect in the platter itself. The thing delaminated physically. Again, long long ago.
     
  14. Palmpilot

    Palmpilot Touchdown! Greaser!

    Joined:
    Apr 1, 2007
    Messages:
    11,366
    Location:
    PUDBY

    Display name:
    Richard Palm
    I once had a drive that had become unusable come back to life when I mounted it in a drive tray that had a cooling fan in it. That was when I started paying attention to temperatures. (Obviously I only used it to the extent required to recover the files from it after that.)
     
  15. denverpilot

    denverpilot Taxi to Parking

    Joined:
    Nov 8, 2009
    Messages:
    32,603
    Location:
    Denver, CO

    Display name:
    DenverPilot
    At one particularly unprepared company I watched the Senior sysadmin bang a drive on the datacenter floor to unstick balky bearings, so it'd run just long enough to copy the system to a replacement machine.

    No, he didn't have a recent backup.
     
  16. Palmpilot

    Palmpilot Touchdown! Greaser!

    Joined:
    Apr 1, 2007
    Messages:
    11,366
    Location:
    PUDBY

    Display name:
    Richard Palm
    At one point, hard drive problems convinced me that I needed to set up an automated backup to an external drive, so I did. The manual backup plan I had before was clearly not being used often enough!