1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Hard Drive Failure Rates

Discussion in 'Storage' started by RCoon, Jan 21, 2014.

  1. RCoon

    RCoon Forum Gypsy

    Joined:
    Apr 19, 2012
    Messages:
    6,318 (7.74/day)
    Thanks Received:
    2,778
    Location:
    Gypsyland, UK
    Rowsol, james888, ne6togadno and 3 others say thanks.
  2. Ahhzz

    Ahhzz

    Joined:
    Feb 27, 2008
    Messages:
    2,066 (0.89/day)
    Thanks Received:
    636
    I knew to stay away from Seagate, and prefer the WD Blacks, but I'll have to look a little more at the Hitachis. Thanks :)
  3. rtwjunkie

    rtwjunkie

    Joined:
    Jul 25, 2008
    Messages:
    1,030 (0.47/day)
    Thanks Received:
    319
    Location:
    Louisiana
    I've still got a couple WD 74Gb Raptors (one in my son's PC, and one in HTPC) that are still in use since Sept 2006. Both still have no SMART warnings, and still have no bad sectors. Simply amazing the longevity of those things!
    Last edited: Jan 21, 2014
  4. FordGT90Concept

    FordGT90Concept "I go fast!1!11!1!"

    Joined:
    Oct 13, 2008
    Messages:
    13,319 (6.34/day)
    Thanks Received:
    3,344
    Location:
    IA, USA
    I have two of the 3 TB Seagates and neither has failed yet. I've kind of written Hitachi off lately but that was clearly a mistake. Bare in mind that Hitachi is the smallest HDD manufacturer out of the three so that may be skewing results.
    Crunching for Team TPU
  5. Arjai

    Arjai

    Joined:
    Apr 3, 2012
    Messages:
    1,843 (2.22/day)
    Thanks Received:
    2,608
    Location:
    St. Paul, MN
    I have two Hitachi's, a 250GB in, This Old Comp, and a 750GB in my ASUS Laptop.

    I the 250 has been stellar! The 750, in my laptop, so far, so good, It is the newer of the two and has been used for a shorter time.

    When I, eventually, get my life back to squared away, I plan to get a Kingston SSD kit and use this 750 for a backup and data drive. Blah, Blah. :lovetpu:

    Thanks @RCoon for the post.
    Crunching for Team TPU
  6. Ahhzz

    Ahhzz

    Joined:
    Feb 27, 2008
    Messages:
    2,066 (0.89/day)
    Thanks Received:
    636
    Actually, click thru to the orig report, you'll see that they had 13000 Seagate and hitachis in the mix, and around 3500 wd drives.
    adulaamin and FordGT90Concept say thanks.
  7. AsRock

    AsRock TPU addict

    Joined:
    Jun 23, 2007
    Messages:
    10,692 (4.15/day)
    Thanks Received:
    1,623
    Location:
    US
    Info not worthy, lack of details.

    How ever i am a WD fan regardless what these kind of reports say.
    Frick says thanks.
  8. FordGT90Concept

    FordGT90Concept "I go fast!1!11!1!"

    Joined:
    Oct 13, 2008
    Messages:
    13,319 (6.34/day)
    Thanks Received:
    3,344
    Location:
    IA, USA
    That's not representative of the market at all. :eek:
    Crunching for Team TPU
  9. micropage7

    micropage7

    Joined:
    Mar 26, 2010
    Messages:
    5,564 (3.54/day)
    Thanks Received:
    1,272
    Location:
    Jakarta, Indonesia
    just saw it from facebook.
    so size matter?
    and i dunno, bigger space can save many data but when suddenly it die, you lose it all
  10. freaksavior

    freaksavior To infinity ... and beyond!

    Joined:
    Dec 11, 2006
    Messages:
    8,047 (2.90/day)
    Thanks Received:
    880
    The 1.5Tb had bad firmware causing them to fail, after about a dozen RMA's I finally got some that worked.

    That's why when you use large file systems you need something to check the data often enough to prevent loss like ZFS. Standard raid levels were not designed for this type of capacity. Now if you're just running one then you can use a backup service or buy a second for backups.
    Crunching for Team TPU
  11. LAN_deRf_HA

    LAN_deRf_HA

    Joined:
    Apr 4, 2008
    Messages:
    4,498 (1.96/day)
    Thanks Received:
    931
    I've never had a HDD outright fail. Had them spit out errors but they all seemed to shape up after a reformat. I switched to Seagate because I got tired of the racket WDs make when seeking. Now I would never go back. Same speed, less noise, lower price.
  12. extrasalty

    Joined:
    May 13, 2009
    Messages:
    176 (0.09/day)
    Thanks Received:
    25
    I was big Seagate fanboi until I had to replace 7(seven) in 1 year- all different ages and sizes. One of the replacements came dead. 1 died a year later. Long turn around and poor refurbishing from Seagate's Tennessee plant. Costly returns- whether 1 or 4 HDDs. Not even factoring their poor manufacturing and quality issues.
    Western Digital is the exact opposite- fast turnaround from California, very cheap shipping - $5 for something UPS charges $11, replacement drives come with nearly full warranty as opposed to refurbs from Seagate, which sometimes have no remaining warranty but 90 days. I made the decision 2 years ago to never buy Seagate ( Samsung too since they are owned by Seagate and warranty is handled exactly the same).
    Another lesson learned - don't use WHS drive duplication. Ever.
    Since all hard drives will fail sooner or later, it gets really hard to make sure your data is not corrupted before that. SMART is not doing much at all for data integrity, let alone warning. After that abomination Windows Home Server drive duplicaton slowly corrupted over 2TB of data, ZFS made me see the light. For me ZFS and WD Black is the winning solution. I laugh at Seagte's peeling magnetic layer now.
  13. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    If you adjust the numbers by only including the numbers you choose to include and purposely leaving others off, you can make it look as bad as you want for any manufacturer you want.

    Every company has a bad run from time to time. I don't prefer any company when it comes to hard drives, I buy what provides the best GB/$, and never rely on a single drive to house any important data. Make backups!
    suraswami says thanks.
    Crunching for Team TPU
  14. Sasqui

    Sasqui

    Joined:
    Dec 6, 2005
    Messages:
    7,430 (2.36/day)
    Thanks Received:
    1,270
    Location:
    Manchester, NH
  15. Ahhzz

    Ahhzz

    Joined:
    Feb 27, 2008
    Messages:
    2,066 (0.89/day)
    Thanks Received:
    636
    Not to disagree with your conclusion by any stretch, but it appears that the only ones they left off, they COMPLETELY left off, due to being completely skewed. they also commented directly to that, saying that the greens were just not dealing well in that environment. Didn't say they were crappy drives or anything. As for a "bad run", if you go back thru their blog, they've obtained these drives from multiple sources, not one set. It's a pretty good read. The data I liked best was the quantity of drives overall. this was not a limited run of even a few hundred drives
    Quote:
    Brand Number of Drives Terabytes Average Age in Years
    Seagate 12,765 39,576 1.4
    Hitachi 12,956 36,078 2.0
    Western Digital 2,838 2,581 2.5


    Toshiba 58 174 0.7
    Samsung 18 18 3.7
    end quote

    sorry, my table-fu is not strong :)

    Looks to me like that's a really wide spread of drives, large collection, and the best Actual Use report I've ever seen on drives. Who else has put that many different brand drives thru that long of a run, and published the details? I have long preferred WD Blacks, have several 2Tb in the house right now, with 2 3Tb reds in my NAS. But this will make me consider the hitachi brand next time out...
    Last edited: Jan 21, 2014
    RCoon says thanks.
  16. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    Yes, but if you are releasing studies on hard drive reliability, if you want to seem reputable don't leave off an entire group of drives just because they all failed and you don't want to show that information.

    Also, I'm not talking about a bad production run, I'm talking about a bad run in terms of a bad stretch of time where they are having issues. Every year we see reports like this, and every year the company with the problems seems to change. A month from now WD could run into issues and their reliability could start to go down.

    Also, one of the 1.5TB Seagate models had an annual failure rate of 120%...yeah their numbers are reliable...:rolleyes:
    Last edited: Jan 21, 2014
    Crunching for Team TPU
  17. Ahhzz

    Ahhzz

    Joined:
    Feb 27, 2008
    Messages:
    2,066 (0.89/day)
    Thanks Received:
    636
    You really should read the whole article. The 120% failure means they replaced them more than once in a year, due to what they assume is a bad batch from refurbs. Again, it doesn't look like they're sugar coating anything except dropping off what appear to be some obvious anomalies, which is what any statistician worth .500000000001 a grain of salt is going to do :)


    Quote:
    We got them from Seagate as warranty replacements for the older drives, and these new drives are dropping like flies. Their average age shows 0.8 years, but since these are warranty replacements, we believe that they are refurbished drives that were returned by other customers and erased, so they already had some usage when we got them.
    End Quote

    And after nagging you to read the article, I missed this gem at the bottom
    "A year and a half ago, Western Digital acquired the Hitachi disk drive business. Will Hitachi drives continue their excellent performance? Will Western Digital bring some of the Hitachi reliability into their consumer-grade drives?" ;)
  18. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    I read the whole article, I just don't agree with their methods. I believe it makes their numbers in-accurate. Something can't fail more than once, if the replacement fails then that is still only a 100% failure rate. The replacement drive counts as a new drive in the study.

    Furthermore, they talk more than once about the fact that they believe a lot of the drives they use are failing because they are using them in environments the drives weren't meant to be used in. These are high-vibration, high temperature, RAID environments. And not surprisingly, the desktop drives that aren't supposed to be used in these environments fail a lot. They are using Seagate desktop drives, which aren't supposed to be used in a RAID environment of more than 2 drives and they are failing. But the WD RED drives that are meant for this type of environment are showing to be more reliable. This isn't a surprise. However, they are presenting this data in a way to get people to believe these are how the drives perform in a desktop environment, or at least they knew that is how people would view the results.

    In fact, if you look at similar studies, such as the one from Google, you see they specifically don't name brand names because they admit that they are using the drives outside of their intended purpose and it would be unfair to use the numbers to influence brand buying decisions of the standard consumer. This is because a drive might behave very poorly in a data center, but will run flawlessly in a desktop.
    Last edited: Jan 21, 2014
    Crunching for Team TPU
  19. FX-GMC

    FX-GMC

    Joined:
    Aug 26, 2013
    Messages:
    824 (2.56/day)
    Thanks Received:
    309
    Location:
    Cumberland Plateau
    I agree.

    Now if they wanted to go into Failures per purchased drive.....
  20. Ahhzz

    Ahhzz

    Joined:
    Feb 27, 2008
    Messages:
    2,066 (0.89/day)
    Thanks Received:
    636
    It seems that you just want to insist that they (Backblaze) produced the results of this study to influence people into buying or not buying a series of drives. This is a long running blog of a company dedicated to "our cloud storage", as referred in the opening paragraph. They had a drive shortage, as many did, they had to source many drives from many different companies to keep up with their demand, and they decided to keep track of the drives replaced, and, in their experience, the drives' real-world MTBF, in a Cloud Environment:
    "In the Backblaze environment..."
    "Some drives just don’t work in the Backblaze environment"

    This is an Online Cloud Storage/Backup company. They have no apparent desire to influence people to purchase one series of drives over another, and in fact, several times stated the have a preference for the WD reds, which they don't portray as having many of at all:
    "Our other favorite is the Western Digital 3TB Red (WD30EFRX)."
    "We wish we had more of the Western Digital Red 3TB drives (WD30EFRX)"
    They do speak highly of the Hitachi drives they have ("If the price were right, we would be buying nothing but Hitachi drives"), but again, people reading this blog are made aware UP FRONT (if for some reason they assumed instead of reading) that this is in a Cloud Storage environment, not a bank of desktop computers.

    Simply put, this blog shows a wide variety of drives in a large databank cloud storage environment, larger than most of us could dream of at this point, but just on a larger scale than what many of us have in our house or place of business. As such, it is highly interesting to me that the WD Red, designed for the RAID environ, scored worse than the equivalent Hitachi Desktar.

    Western Digital Red
    (WD30EFRX)3.0TB __________346_______________0.5_______________3.2%
    Hitachi Deskstar 7K3000
    (HDS723030ALA640)3.0TB ___1027_____________2.1________________0.9%


    As for " the replacement fails then that is still only a 100% failure rate.", that's just playing with numbers. As a computer tech in this forum, I had absolutely no problem understanding that that meant they had drives fail and be replaced in less than a year's time. If someone didn't understand that, they were reading the wrong blog, and meant to be reading about Paddington Bear, not Backblaze. (Plainly, how would you explain that they bought **for instance** 100 drives, and had 120 fail in one year, on average? I understood it...)

    What's the issue here? It almost sounds like a fanboi rant from you, which I'm sure you aren't. The company provided some serious real-world data, description of the environment used, reasoning behind the layout and build structure, and their experience with several different drive types. They appear to have no hidden agenda, no company affiliation beyond their own, I don't see any harm done that anyone who cares to read these type articles could possibly experience. If Joe Public decides to read it, and then decides to buy a Hitachi over a Western Digital or Seagate, who is harmed? If it's a question of "Seagate's not that bad!!!", Tom's Hardware readers appear to disagree, and if any of us are going to take someone's word over another, what does it matter that there's another dissenting opinion?

    To the educated reader, all the information explaining the parameters of the "test" are in plain text. To the uneducated, they're still wondering if their microwave is spying on them for the NSA.....
    Last edited: Jan 22, 2014
    RCoon says thanks.
  21. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    I think I've made myself very clear. This isn't a fanboy rant, I'm just stating my opinion.

    Measuring drive failures at more than 100% makes the numbers inaccurate in my opinion(and it seems I'm not the only one). And in fact in statistical studies as well. In medial studies, if the patient is cured, then relapses and is cured again they don't call that a 200% success rate. If a patient dies, and is revived, and then dies again it isn't a 200% failure rate. To be really accurate, the new drives, even if they were replacements for failed drives, should be counted as a separate drive.

    Also, releasing the brands of the drives only leads to people taking the study's information incorrectly, and why most companies that release numbers like this don't release the specific brands of the drives. The information is interesting for sure. However, to release the brand names only serves to try to influence people's buy decisions. And just as expected you have sites like TechReport and others picking these numbers up and presenting them as if they were desktop failure rates, and people here believing that. You can't use these numbers to predict desktop failure rates. These were drives put into conditions they weren't meant for, they would likely have very different failure rates in a desktop system.

    And the Hitachi Deskstar drives are very well built drives with rock solid firmware. Hitachi actually approves them for RAID use. So it isn't surprising they are doing well and even outperforming the RED drives(which are just Green drives with TLER enabled). But that is the point too. Hitachi doesn't really have an enterprise class drive, their standard desktop drive fills that niche too. So they write their firmware to be RAID compatible. Seagate and WD on the other hand don't write the firmware for their desktop drives to be RAID compatible. But remember, there was a time when the Deskstars weren't that great, which is why they earned the nickname Hitachi Deathstars.
    Crunching for Team TPU
  22. Ahhzz

    Ahhzz

    Joined:
    Feb 27, 2008
    Messages:
    2,066 (0.89/day)
    Thanks Received:
    636
    Homer: "I agree with you in theory...."

    So, patient recovery rates aside, how would you indicate that you purchased 100 drives, and over the course of a year, had 120 drive failures? I think they explained their numbers sufficiently for anyone to understand.

    I don't think either one of us can speak to "only leads to people taking the study's information correctly....". I understood it. You understood it. Several people here understood it. We can't correct everyone :)
    [​IMG]

    I absolutely remember the Deathstars, well old enough for that, and remembering the IBM tags on hard drives, and several other defunct names, *shudder* maxtor included. I also didn't know that about the greens and reds. I had read some about them back when I got my baby NAS over here, I think, but didn't see that info. Checking Freenas right now for more info, thanks :)
  23. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    It is simple, if the 100 drives they purchased died, then it is a 100% failure rate. If the drives are replaced, then that is a new drive in the study. So if they had 120 drives fail, then the had 120 drives. You can't have 120 drive fail if you didn't have 120 drives.

    The fact that we've already seen techreport mis-reporting the information, and people here talking about the information like it applies to desktop applications. Yes, if you read the whole study you get all the information, but most people won't, most people won't even read the entire TechReport article. Most people will look at the pretty graph and that is it.
    Crunching for Team TPU
  24. larrymoencurly

    Joined:
    Apr 21, 2007
    Messages:
    23 (0.01/day)
    Thanks Received:
    4
    Are there any explanations for the different failure rates?

    Russia's StoreLab agreed about Hitachis being best but thought post 7200.11 model Seagates were fine.
  25. RCoon

    RCoon Forum Gypsy

    Joined:
    Apr 19, 2012
    Messages:
    6,318 (7.74/day)
    Thanks Received:
    2,778
    Location:
    Gypsyland, UK
    Then in that case they deserve to be horribly misinformed and advance absolutely nowhere in their technical adventures.
    I find all the information useful, regardless of how confusing the case study was carried out. Somebody has finally given us a useful report to base our HDD failure complaints on, and specifically outlines certain drives that cannot hack it when it comes to heavy duty data storage use.
    All in all, GG Hitachi.
    Ahhzz says thanks.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page