Hard Drive Reliability Review for 2015

By | February 16th, 2016

Backblaze Hard Drive Stats
By the end of 2015, the Backblaze datacenter had 56,224 spinning hard drives containing customer data. These hard drives reside in 1,249 Backblaze Storage Pods. By comparison 2015 began with 39,690 drives running in 882 Storage Pods. We added 65 Petabytes of storage in 2015 give or take a Petabyte or two. Not only was 2015 a year of growth, it was also a year of drive upgrades and replacements. Let’s start with the current state of the hard drives in our datacenter as of the end of 2015 and then dig into the rest later on.

Hard Drive Statistics for 2015

The table below contains the statistics for the 18 different models of hard drives in our datacenter as of 31 December 2015. These are the hard drives used in our Storage Pods to store customer data. The Failure Rates and Confidence Intervals are cumulative from Q2 2013 through Q4 2015. The Drive Count is the number of drives reporting as operational on 31 December 2015.
Hard Drive Relability 2015

During 2015, five drive models were retired and removed from service. These are listed below. The Cumulative Failure Rate is based on data from Q2 2013 through the date when the last drive was removed from service.

Hard Drives Removed 2015

Note that drives retired and removed from service are not marked as “failed” they just stop accumulating drive-hours when they are removed.

Computing Drive Failure Rates

This is a good point to review how we compute our drive failure rates. There are two different ways to do this, either works.

For the first way, there are four things required:

  1. A defined group to observe, in our case a group of drives, usually by model,
  2. a period of observation, typically a year,
  3. the number of drive failures in the defined group over the period of observation, and
  4. the number of hours a group of drives are in operation over the period of observation.

Let’s use the example of 100 drives which over the course of 2015 accumulated a total of 750,000 Power-on-hours based on their SMART 9 RAW values. During 2015, our period of observation, five (5) drives failed. We use the following formula to compute the failure rate:

(100*drive-failures)/(drive-hours/24/365)

(100*5)/(750,000/24/365)

This gives us a 5.84% annual failure rate for 2015.

For the second method, the only change is how we count the time a group of drives is in service. For a given drive we simply count the number of days that drive shows up in our daily log files. Each day a drive is listed it counts as one drive-day for that hard drive. When a drive fails it is removed from the list and its final count is used to compute the total number of drive-days for all the drives being observed.

Drives by Manufacturer

The drives in the datacenter come from 4 manufacturers. The following chart shows the cumulative hard drive failure rates by manufacturer for all drives:
Hard Drive Failures by Manufacturer
Let’s take a look at the “make up” of the drives in our datacenter.

Hard Drive Count 2015 2015-drive-days-piechart

The chart on the left is the total number of drives and the chart on the right is the total number of drive hours for all the data drives by each manufacturer. Notice there are more Seagate drives but the HGST drives have more hours. The HGST drives are older, as such they have more drives hours, but most of our recent drive purchases have been Seagate drives. Case in point, nearly all of the 16,000+ drives purchased in 2015 have been Seagate drives. Of the Seagate drives purchased in 2015, over 85% were 4TB Seagate drives.

Hard Drive Reliability by Drive Size

1TB Hard Drives

We removed the last of our 1TB drives during Q4 and ended the quarter with zero installed. This was done to increase the amount of storage in a 1TB filled Storage Pod as we replaced the 1TB drives with 4TB drives (and sometimes 6TB drives). Now in the same Storage Pod we get four times as much data. The 1TB Western Digital drives performed well with many of the drives exceeding 6 years in service and a handful reaching 7 years before we replaced them. The cumulative annual failure rate was 5.74% in our environment, a solid performance.

We actually didn’t retire these 1TB WD drives – they just changed jobs. We now use many of them to “burn-in” Storage Pods once they are done being assembled. The 1TB size means the process runs quickly, but is still thorough. The burn-in process pounds the drives with reads and writes to exercise all the components of the system. In many ways this is much more taxing on the drives then life in an operational Storage Pod. Once the “burn-in” process is complete, the WD 1TB drives are removed and we put 4- or 6TB drives in the pods for the cushy job of storing customer data. On the other hand, the workhorse 1TB WD drives are returned to the shelf where they dutifully await the next “burn-in” session.

2TB Hard Drives

The Seagate 2TB drives were also removed from service in 2015. While their cumulative failure rate was slightly high at 10.1%, they were removed from service because we only had 225 of those drives and it was easier to upgrade the Storage Pods to 4TB drives than to buy and stock the 2TB Seagate drives.

On the other hand, we still have over 4,500 HGST 2TB drives in operation. Their average age is nearly 5 years (58.6 months) and their cumulative failure rate is a meager 1.55%. At some point we will want to upgrade the 100 Storage Pods they occupy to 4- or 6TB drives, but for now the 2TB HSGT drives are performing very well.

3TB Hard Drives

The last of the 3TB Seagate drives were removed from service in the datacenter during 2015. Below is a chart of all of our 3TB drives that were in our datacenter anytime from April 2013 through the end of Q4 2015.

3TB Drive Review

4TB Hard Drives

As of the end of 2015, 75% of the hard drives in use in our datacenter were 4TB in size. That represents 42,301 drives broken down as follows by manufacturer:

4TB Drive Stats

The cumulative failure rates of the 4TB drives to date are shown below:

4TB Hard Drive Reliability

All of the 4TB drives have acceptable failure rates, but we’ve purchased primarily Seagate drives. Why? The HGST 4TB drives, while showing exceptionally low failure rates, are no longer available having been replaced with higher priced, higher performing models. The readily available and highly competitive price of the Seagate 4TB drives, along with their solid performance and respectable failure rates, have made them our drive of choice.

A relevant observation from our Operations team on the Seagate drives is that they generally signal their impending failure via their SMART stats. Since we monitor several SMART stats, we are often warned of trouble before a pending failure and can take appropriate action. Drive failures from the other manufacturers appear to be less predictable via SMART stats.

6TB Hard Drives

We continued to add 6TB drives over the course of 2015, bringing the total to nearly 2,400 drives (1,882 Seagate, 485 Western Digital.) Below are the Q4 and Cumulative Failure Rates for each of these drives.

6TB Drive Stats

The Seagate 6TB drives are performing very well, even better than the 4TB Seagate drives. So why do we continue to buy 4TB drives when quality 6TB drives are available? Three reasons:

  1. Based on current street prices, the cost per TB of the 4TB drives (0.028/GB) is less that of the 6TB drives (0.044/GB).
  2. The channels we purchase from are not flush with 6TB drives, often limiting sales to 50 or 100 units. There was a time during our drive farming days when we would order 50 drives and be happy, but in 2015 we purchased over 16,000 new drives. The time and effort of purchasing small lots of drives doesn’t make sense when we can purchase 5,000 4TB Seagate drives in one transaction.
  3. The 6TB drives like electricity. The “Average Operating Power” is 9.0W for a Seagate 6TB drive. That is 60% more than the 5.6W used by the 4TB Seagate drives we use. When you have a fixed amount of power per rack, this can be a problem. The easy answer would seem to be to add more electric to the rack, but as anyone who designs datacenters knows it’s not that simple. Today, we mix 6TB filled Storage Pods and 4TB filled Storage Pods in the same rack to optimize both power consumption and the storage space per square foot.

5TB and 8TB Hard Drives

We continue to only have 45 of each of the 5TB Toshiba and 8TB HGST Helium drives. One 8TB HGST drive failed during Q4 of 2015. Over their lifetime the 5TB Toshiba drives have a 2.70% annual failure rate with 1 drive failure and the 8TB HGST drives have a 4.90% annual failure rate with 2 drive failures. In either case, there is not enough data to reach any conclusions about failure rates of these drives in our environment.

Drive Stats Data

Each day we record and store the drive statistics on every drive in our datacenter. This includes the operational status of the drive, along with the SMART statistics reported by each drive. We use the data collected to produce our Drive Stats reviews. We also take this raw data and make it freely available by publishing it on our website. We’ll be uploading the Q4 2015 data in the next few days then you will be able to download the data so you can reproduce our results or you can dig in and see what other observations you can find. Let us know if you find anything interesting.

Andy Klein

Andy Klein

Andy has 20+ years experience in technology marketing. He has shared his expertise in computer security and data backup at the Federal Trade Commission, Rootstech, RSA and over 100 other events. His current passion is to get everyone to back up their data before it's too late.
  • Rachelle

    Forgive my ignorance, as I am not a computer expert and I’m feeling overwhelmed by all the information out there. I have a Dell Ultrabook 15z that was about 4 months shy of its 4th year when the hard drive began failing. I was able to back up everything in time. I’m trying to figure out the best replacement hard drive. “Best” in the sense of biggest bang for my buck with speed and longevity. being most important to me. I’m not a high-end gamer, and primarily use it for internet, photo/video/music storage, Word/Excel, and simple hidden object or match 3 games, pretty much in that order. I’d appreciate anyone’s feedback/suggestions.

    • You’re probably not going to glean the data you’re looking for from this post – the drives we use are 3.5-inch models you’d find in a desktop computer as opposed to the smaller 2.5-inch ones you need in a laptop.

      Having said that, I’ve found “best bang for the buck” when it comes to storage performance for older laptops is to replace the spinning hard disk drive with an solid state drive (SSD). SSDs offer much faster read and write speeds than hard drives do, don’t use as much power and don’t make noise, either. You’ll see your computer booting faster, apps will load quicker, and the system will just in general be more responsive with an SSD than with an HDD.

      Lots of companies make SSDs that are plug-and-play replacements for hard drives.

      In terms of longevity, SSDs are better than ever before. But the most important thing is that you don’t put all your eggs in one basket – SSD or hard drive, having a solid backup strategy is vital. That way, when your machine inevitably develops a problem, you’l have something to recover from.

      • Rachelle

        Thank you for your feedback. It was helpful :)

  • well we cant judge the manufacture reliability by this mere post

  • MrRandom

    I am surprised to see WD so close the Seagate when it comes to failure rate… In my experience over the years I have had so many drives from Seagate fail that I refuse to buy them.

  • Pillion Mac

    Hitachi Hard Drives:
    ===============
    It is said that data is the soul of your business. And without a data storage medium, the soul cannot flourish. Therefore, reliable storage mediums, such as the Hitachi Hard Drives, are in high demand that can keep the data safe and stable. They are particularly popular for their low failing ratio.

    However, some customers have no idea about how and where to buy these devices.

    Keep data safe and sound
    This is where Memory4less.com steps in. Being the top-quality solutions provider for data centers, it enhances your data capacity by providing estimates and consultancy for large data centers.

    For personal and commercial use
    We, at Memory4less.com, pride ourselves in providing exceptional storage capacity and robust speed for both homes and offices.

    Browse the Hitachi Storage medium and Hard drives below to find your best-fit. Simply click on the link to get a quote.

    Still unsure about the storage medium? Get in touch with our Memory professionals to discuss your dilemma – you will definitely find a solution here.

    http://www.memory4less.com/hitachi-hard-disk-drives

    • Harry B. Furr

      shill much?

  • I keep seeing seagate getting a lot of crap, but here it looks like WD have the larger failure rate? (At least so far in Q1 2016). I’m now wondering what to do as I’d kinda decided on seagate because I want SSHD drives (Hybrids). And it seems they are the only company offering 4TB with 7200RPM platter speed. Unless anybody can suggest otherwise?

    • Aitor Bleda

      Why use SSHD? it is only marginally better than a regular HDD… the only use I see for them is a games console. And I do own one of them.

  • Ian Stevens

    Am I missing something? Everyone is slamming Seagate, but the overall failure rate by manufacturer showed WD to be significantly worse than the other three.

  • Babz

    Does a drive last longer when it sits as an archive than when it is in an enclosure with the power on? I would like to have a cloud service, a hard copy and a second copy at a different location. I need 14-20 TB of storage per copy.

  • Hmmmm Pretty Interesting statistics !!! Impressed

  • johndavidwright

    Thanks for posting these blog posts!

  • Stefan

    How about collecting and adding statistics of the number of DOA’s? (Dead on Arrival)

  • Chris Parkin

    Although you monitor SMART stats do you also run SMART tests on your drives? If so short & long & how often? If so how do you schedule this to avoid a performance hit? BTW LOVE the blog :)

  • John Barksdale

    Mr. Klein thank you for this report. As a system administrator I enjoy reading your reliability reports. Excellent writing!

  • JH

    ITT: OMG I have had 3 Seagate/WD/HGST/Toshiba hard drives and they all failed!!! I am commenting on a report containing literally tens of thousands of data points, but I am going to ignore all of them and provide my own expert opinion because my vast personal experience makes me absolutely sure that Seagate/WD/HGST/Toshiba makes the worst drives ever!!!!!

    • Parsifal Druddle

      oh man, thanks for that =)

  • Raymond Savoie

    Those 29,000+ Seagate 4TB drives (ST4000DM000) they are desktop-class drives and are not NAS certified by Seagate, unless I’m mistaken. Should that be a factor?

    Are Seagate NAS drives just marketing fluff? Does Backblaze re-flash the drives with a NSA-specific customization, or perform some parameter tweaks to get them to behave nicely in a NAS environment? Or maybe the way they are deployed and used at Backblaze doesn’t require NAS-specific drive features? Given how many were purchased it could also be self-insurance?

    Fascinating stuff, either way. Thanks for the article!

    • leexgx

      no NAS firmware was used they are just consumer drives in 60-80 drive pods (really at the time only WD and HGST would of been best choice due to secured disk at both ends)

      the 3-4TB drives are just pure bad

      the WD and HGST i believe the disk is secured at Both ends whereas seagates are only secured at one end so they are problematic to vibration especially the ST4000DM000 and ST3000DM000 that had a high fail rate of 25% (somthing like that) somthing they fixed in there newer drives (and did not affect their older drives)

      my ST3000DM000 started to cascade fail on me in reallocated sectors death but in a nice slow way, luckily i had 0 uncorrectable errors but the the reallocated sectors was mounting up by the time i had copied all the data off it (to a WD red 1st gen 4TB) the health hit 10% and was setting the windows backup your data warning popup box (as well as making my smart tool go bonkers at me way before that)

      once you see reallocated sectors you should change a HDD right away and uncorrectable errors as well, SSD not so much unless its a cascading fail where your SSD is continually having more relocations (i have a number of SSDs that have had relocations but it was just a one time event no more have appeared in 6 months or they came with relocated events on older SSDs which was normal at the time for SSDs)

  • Tommy

    I always find these posts very interesting to read, but in as far as deriving any conclusions from this as to which manufacturer’s drives to buy for yourself, there is really only one thing that stands out – HGST drives are reliable. The question is then if the extra cost is worth it to you. Otherwise the rest of the stats aren’t truly relevant unless you are buying the exact model of drives they have in service. It is also interesting to note that while they don’t have many of them, the WD Red drives aren’t as reliable as I would expect.

    I understand the reasons for buying a small variety of drives in bulk, but I wish they would at least buy a few pods worth of other models for comparison’s sake. For example, I’m curious how the WD blue and black drives would perform in this environment even though they supposedly don’t have the specific attributes the red drives have for a NAS enclosure. I.E. do those design differences actually make a difference? I’ve always been skeptical.

  • Trejkaz

    “Today, we mix 6TB filled Storage Pods and 4TB filled Storage Pods in the same rack to optimize both power consumption and the storage space per square foot”

    Nit-picking: If you want to optimise storage space per square foot, you don’t want to waste your time with 4TB drives in there. If you mean that you put 4TB drives in there before it wasn’t possible to run the 6TB ones due to a lack of power, then what you mean is that you made a compromise, not that you optimised both those things, because clearly storage space per square foot is not optimised.

    • EssJ

      If they filled the pod with 6TB exclusively and the power draw exceeded capacity, then that would not be an optimal configuration as the pod would be dead.

      My interpretation is that they put as many 6TB drives as they could without exceeding the maximum power capacity, which would optimize both metrics.

  • OllieJones

    Question: the HGST drives with best reliability have low power modes, according to their data sheets. Do your Storage Pods allow them to spin down? Do you have any idea what the spun-up / spun-down duty cycle is on those drives?

    Thanks for your always fascinating writeup.

  • DanThePhotoMan

    It looks to me that HGST stands above the crowd for reliability. Am I missing something here?

    • Rick Baumhauer

      Nope – if your budget supports it, HGST is generally the best choice for reliability, and has been for some time.

      The big issue (especially for a volume purchaser like Backblaze) is cost – HGST drives are always expensive, especially when they discontinue their fantastic 4TB drives instead of gradually lowering the price over time as the other manufacturers do. 4TB is the sweet spot in cost/GB of storage right now, but HGST (like Apple, in a way) goes after the high profit part of the market, and 4TB drives have moved to “commodity” status, so they’ve moved on to the high margin 6-8TB drives. Their parent company handles the lower end consumer drives and some specialized markets (Black for performance, etc).

  • Ross Lazarus

    There’s a more complete description of my KM analysis of the raw data in case anyone cares at http://bioinformare.blogspot.com.au/2016/02/survival-analysis-of-hard-disk-drive.html .
    I’m happy to help if anyone wants to adapt the method to provide better insight into the data?

  • D’ohrk!

    Why do you suppose, with WDC owning HGST, that WDC’s own drives suffer poorly in performance, by comparison?

  • Chris Minasians

    Fantastic tests, thanks again! Really interesting to read these :)
    I’m still using my Seagate Baracuda 3TB for over 4 years now!

  • Marek Grochocki

    So if you RMA failed drives do you get and use recertified/refurbished drives?
    Is there a statistic for refurbished drives? It would be nice to know.

  • RH

    Never had a problem with WD, even in a laptop that gets bounced around daily. Maybe they sell MORE and that’s why the failure rate is higher. The HGST drives are nice, but bang for the buck is what I look for.

    • EssJ

      Volume would have no effect on rate.

  • Calvin Hilton

    Andy, given the failure rate variation within a single manufacturer how much weight would you give your findings to drives you don’t use made by the same manufacturers?

    I’ve been using WD Black in my home systems for many years without any problems. My most recent purchases have been 4TB and 6TB drives. Should I switch to a different manufacturer based on your findings? Thanks.

    • twitchyfirefly

      This is why I view these BackBlaze reports as interesting but by no means definitive. Being a consultant for almost 20 years I’ve bought and installed a lot of hard drives; my current favorite drive is WD Black. The WD Blue and Green have failed too many times for me, but I’ve had good luck with Black.

      • EssJ

        Unless you buy and use 16,000 drives a year as a consultant, I’ll put more credence in their hard data versus your anecdotal experience.

  • Reisk

    I see people voicing a dislike for Seagate drives. From my experience I agree, and for my personal use I usually buy Western Digital. But I believe that the data offered here is correct, and it’s consistent with what I’ve observed in my machine rooms. The caveat is that Seagate drives aren’t as dependable when they’re allowed to power down and cool off. A few years back we bought a large drive array for a major chron system, and I was appalled when the vendor supplied us with Seagate drives. But they were fine. Months later we were discussing repopulating an older array and one of the server monkeys mentioned buying Seagates. I voiced my dislike of Seagates, but he grinned and said, “Seagates are good as long as you thrash them.” His opinion was that you don’t want Seagates in systems that would regularly go into sleep, and I found out that for those systems he spec’d Toshibas.

  • Logic First

    TLDR for those who didn’t read the article: Backblaze torture tests only WD hard drives and then people are surprised when the WD’s have a higher failure rate. Considering your past tests showed Seagate with a horrible failure rate, then you choose to only buy seagate drives and subsequently trash the reputation of WD by “pounding them” and counting them as failures, I guess my only question is “how much is Seagate paying you”?

    • Had Enough With the BS

      Interesting angle.

      • Milk Manson

        You like your angles obtuse, I take it.

        • Had Enough With the BS

          Whatever

    • Andy Klein

      The WD pounding described was for the 1TB WD drives we had removed from service and replaced with larger drives. After a drive is removed from service we don’t count any of the stats that may occur. In this case the WD drives were used to load test systems during the burn in process, they were not placed back in production and the stats produced during this phase of their lives was not included in the results. The numbers represent only time spent in production level Storage Pods.

  • Black705

    Running a Maxtor 6L160M0 about +10 years old, Power-On Hours/Cycle Count Data = 45559 hours / 5.20 years)

    [0C] Power Cycle Count: 238/Always OK, Worst: 238 (Data = 6134). Currently slowing down pc heavily.

  • Thomas Wolf Tompkins

    Interesting since almost every toshiba I have seen has failed with bad sectors also seagate most fail under a year with bad sectors. I have 10 WDC’s in my server bought in 2002 and there still working till this day no issues at all.

  • Chris

    First off, based on this I would be buying HGST drives exclusively. 1.8% or less failure rate? Yes please! I’ve been buying HGST drives (mostly Ultrastar’s) for a couple of years now. They are super fast and reliable in my home NAS. My second choice would be WD RE drives.

    In bulk the price difference between HGST and Seagate cannot be THAT much… I would think the additional reliability would give you a better ROI instead of keeping replacing cheaper drives.

  • FollowTheORI

    Tank you again and keep these posting anytime as possible!

    Keep up the good work!

    Thank you!

  • Rman87

    Based on my experience, I dont see the seagates fail till more like the 3 year point and by four is when you see them dropping like flies. The data shared here is interesting but my experience with seagate has been so bad over the course of a long number of years that i will never ever trust them again.

  • Keyboard_Warrior

    Well this is not much of a news to be honest.
    What is makes me kind of wonder is that, how WD become the worst HDD manufacturer from the best. 15 years ago they were on the top, if you had a WD you were the King and you would laugh on the Maxtor and Seagate owners. Now-days it changed very much, having a WD is kind of a joke. Especially the WD’s Green series! The best thing is stay away from those.
    Although having the RED series at my workplace, I’d say they are reasonable and work just fine, but Green and Blue aren’t good I think…

  • Marcosa

    Why would you post an article with gray text? Black text on a white background works better.

  • biryani

    Has there been any significant change to the way drives are sourced? Someone suggested on another forum that Backblaze was previously buying Seagates off-the-shelf, but that you’ve since switched to sourcing them from Seagate directly, and this has influenced the results.

    • Andy Klein

      We buy drives via the reseller channel. During the Thailand drive crisis when the reseller channel got very expensive we supplemented our purchases with drives we purchased from the consumer channel. Generally, the same drives, different source. We go into that topic in this post: https://www.backblaze.com/blog/3tb-hard-drive-failure.

  • Charles Borner

    So, the basic reason to use Seagate drives are:

    A) They’re cheap
    B) They scream really loud just before they die, hopefully when someone’s listening.
    C) They’re cheap.

    In that case, I’ll just stick with WDC and HGST.

    Sure, they may give up the ghost silently. But if they’re lasting that long in your sweatbox environment, the chances of them spontaneously failing under normal desktop usage is infinitesimally small.

  • Zaphod

    I have a question about how/if your hard drives are ‘maintained’: As part of normal processes, are firmware updates done to your disks, or do they stay as they came out of the box?

    • Andy Klein

      It depends on the nature of the update and how it could apply to us. We can update the drives if needed.

  • HH

    Guys – if you are analyzing data on how long drives “survive”, why don’t you do a standard survival analysis approach? Its stats 101 for clinicians/biologists. And then you wouldn’t have to do those misleading Apples to Oranges comparison between different manufacturers whose drives had a completely different age structure. Cumulative failure rates make little sense in the context of varying ages of drives.

    • Andy Klein

      Over the several different drive stats posts we’ve done we used several different ways to present the data, although we haven’t published a survival analysis (maybe reliability analysis is more correct here) to date. We’ve done something like a lifetime distribution function that showed to the failure rates over time, but that was for all drives, not a specific model. Good food for thought. Thanks.

      • HH

        Very glad to hear. Yes – reliability is certainly the better term – and I am sure there are a few methods I have never heard of that are better suited than the vanilla methods for your type of data.
        In the end, such a survival analysis (the vanilla cox-model) would give you multipliers of failure rates for different drive models, taking age and even replacement without failure appropriately into account. Good Luck.

      • Ross Lazarus

        kaplan meier curves for the 31 million or so rows by manufacturer and model at http://imgur.com/a/rH6Ot if anyone cares? Generated in R after some python processing of the raw csv.

        I think they give interesting details you just don’t get from failure rates. These ignore hours – just take time under observation and deal with right censoring properly so you can distinguish early and late failure patterns….

        Oh, and if you’re wondering about the bogus manufacturer “ST400LM012”, it comes from (eg) 2014/2014-09-26.csv as 2014-09-26,S2ZYJ9DDB27222,ST500LM012 HN,500107862016,0 – this method helps reveal duff data too!
        :)

        • amadvance

          Thanks Ross,

          Very interesting analysis!

  • Dave

    You are hack, HGST are owned by WD and marketed for the low Enterprise level space. WD EFRX are entry level consumer drives. Both made by the same company. Please stop embarrassing yourself like the incompetent that you are.

    I would never purchase any of the services your company offers as it is evident that much like yourself they are incompetent as well.

    • Emmanuel Goldstein

      I guess you didn’t actually read the post.

      • Dave

        Congratulations you made it as retard #2, right after this hack of a company. I would never entrust you or the idiots that published this article to run a proper data center.

        • Emmanuel Goldstein

          As long as it doesn’t effect availability of client data, why does it matter.

          • Steve Hughes

            Agreed. It’s about cost/benefit. I have a small cloud business with about 20RU used for storage. We use the HGST drives and I love the fact that during the last six years we have dropped only one drive. Some of the drives have been spinning 24/7 for the entire six years. For us as a small operation targeting enterprise grade availability it is a major nuisance to be replacing drives, so for me its HGST all the way. But at the other end of the scale it’s obviously more cost effective to dedicate resources to managing cheaper drives wiith a shorter life cycle. There’s nothing hack about that as long as the system design maintains data availability.

          • Dave

            Cost.

    • silvestris

      Google does the same. Can I assume you will be boycotting them for their incompetence as well?

      • Dave

        I would never use the Google data center for Enterprise level software. Azure is the way to go. There you can actually pick the type of drives you want to use.

        • Milk Manson

          Enterprise level software…. you’re such a poser.

    • Aitor Bleda

      You are just trolling.
      Most big scale operations don’t use “Enterprise level” HW because it makes no financial sense to do so. It is far more effective to use consumer level hardware. Most big scale operations have been using consumer level HDDs, they just use smart software.

      If I ran my own datacenter, I would do the same. As I run it for others, I use and demand Enterprise level hardware.

      • Chris

        I disagree. I think it absolutely makes financial sense. The last HP storage array I bought used Seagate ES.2 (enterprise drives). A friend of mine who has experience with EMC SAN’s says when they have ordered drive trays they have came with ES.2 or WD RE drives. Again, enterprise level. I would much rather pay a 10% premium for a drive that lasts at least twice as long. Maybe they cannot get them in bulk, but I find that hard to believe.

        Enterprise drives have better warranties, they are built better, higher performance and are absolutely worth the 10% premium. If they weren’t HP, Dell, EMC, NetApp wouldn’t use them in their products.

        • EssJ

          There’s no indication that enterprise drives last “twice as long”.

          You have to consider the useful life of drives, even ones that never fail. A cloud storage business is always going to feel pressure to increase the amount of storage they offer for the same price. With a given amount of space for racks how are you going to do accommodate that? You have to increase the data density of the racks you have. They seem to do this by systematically replace smaller drives with bigger ones. So, many drives will be retired before they ever fail, even at consumer-level reliability. To use enterprise drives in this scenario, and pay the associated price premium, would be foolish.

      • Dave

        Incorrect, enterprise level hardware not only does it have longer warranties they are designed to be always on devices. Also the price gap between a 7200K RPM entry level Enterprise HDD is not that much more than then the consumer counter part. You actually end up saving money using Enterprise devices. To to lower replacement costs and longer warranty periods.

    • Milk Manson

      If stupid could fly, you’d be a jet.

      • Dave

        If murder were legal, you’d already be dead.

  • Andy Klein

    Several of you have asked questions related to drives failures, how we measure them etc. This is covered in a previous post: https://www.backblaze.com/blog/hard-drive-reliability-stats-for-q2-2015.
    Basically, if it stops working or tells us it is in trouble via SMART stats we pull it.

    • Oracles

      Honestly, Andy, you should prefix and suffix all of these blog entries with “most of your questions can be answered easily by browsing previous posts on this subject.” Every edition of the Reliability Review is chock full of the same “why don’t you…” and “why did you…” questions that you’ve already answered a dozen (or more) times. Maybe just build a table of links pointing to the previous entries that you post right up front each quarter.

  • kevin

    How about graphs by drive model with months of service before failure? Average months of service is rather meaningless. When you see high fail rates, do you ask the vendor for failure analysis info? It appears you are still using desktop drives in a enterprise application, technically that voids the warranty from WD and Seagate. Are you having any issues getting warranty coverage?

    • Andy Klein

      Different subjects, let’s see if I get them all.
      1) Most drive models have a drive or two fail out of the box or during initial testing, so I’m not sure that’s what you are asking for. If you want a timeline of failure points, you’d have to look through the data here: https://www.backblaze.com/hard-drive-test-data.html to build that.
      2) We talk to the vendors about all kinds of things related to their products.
      3) If the vendor believes we fall under warranty they cover us as appropriate.

  • Chris Lutsic

    Is there a reason why you chose to use Seagate “Desktop” models of HDD? I would have assumed your most reliable hard drives would have been the NAS series of HDD. Please let me know as I am intrigued in seeing why you chose this route. Also I am building an 8 bay NAS for media streaming and file shares. Thanks in advance for your response!

    • Andy Klein

      Our environment is built to deal with disk failure. NAS drives are nice, but they don’t perform much (if any) different in our environment and generally aren’t worth the premium. In your case, you’re environment most likely less tolerant of drive failure and so using purpose built NAS drives may be a better choice for you.

    • Emmanuel Goldstein

      Because they pick drives from a cost benefit analysis. It’s cheaper to buy desktop drives and replace them than pay the premium.

      • Andy Klein

        Even better answer +1

        • mickrussom

          Do the manufacturers try and deny warranty claims because you are using a “non-24-hours” desktop drive in a server-storage role. Thinking about the horrific WD30EFRX.

          • Emmanuel Goldstein

            That’s a good question, I wonder if they even bother and just toss the drive. Since most of the time for a warranty RMA, they’ll send you a refurb anyways. So might was well just replace with a new drive.

    • Jesper Monsted

      NAS drives are marginally different from the desktop ones, usually only a different firmware that responds to errors in ways that don’t break RAID as easily. For true high performance and high reliability, it’s the “enterprise” drives (usually with SAS controllers) you’re looking for – but at at least twice the price.

      • Milk Manson

        Twice the price? I wish.

  • Jelly J

    Just bought a 3TB WD Red to replace a 12-month old 3TB WD Green that died during moderate usage. These stats make for depressing reading for me :/

  • Tom Ehlert

    could you give some details how these drives fail?
    do they start to have some bad sectors,
    have one or more SMART parameter beyond some threshold, or are the drives completely dead?

  • Scott Pam

    So far Seagate has the highest amount of crapped out drives in my experience.

    there is a 4TB paperweight under my desk that was just one month past it’s year warranty and went from 100% health to 3% in one day…and all of my photography is on it.

    The iMac 27″ has a failing seagate drive that I can’t even afford to replace yet.

    I have stacks of seagates piled high that I removed from my computer client’s computers.

    Won’t take a seagate for free.

    • jameskatt

      Interestingly despite their reputation WD drives have about the same failure rate as Seagate. But Seagate is so much less expensive that I can easily buy several and have multiple backups. So even if one drive craps out, I have all the other backups available on drives that haven’t yet crashed.

      So I treat hard drives like floppy disks – fragile containers for which it is best to have multiple clones Seagate makes it easier to have multple clones by being so cheap.

    • Rman87

      Thats been my experience as well and this was from 500mb through 3tb versions. I now have ZERO Seagates in operation online and only a few left as offline backup in the safe. The real failure rate from my experience is 3 years plus with by year 4 you have seen most die. Personally these individual posts trying to claim WD has had failure rates as high as seagate makes me question if these are seagate employees. My experience and most people i know that deal with this stuff have experienced what i have. My seagate failure rates have been so high (150%+ because RMAd drives fail again) that i dont want them if theyre free. I have lost so much money as an end user consumer having to replace failed seagate drives that i simply refuse to ever buy them again. I want 8tb archive drives badly but still refuse to buy the seagate. Theres a reason why seagate is always 10% cheaper than WD, think about it.

      • Larry Gall

        I feel the same way. I read this with an odd look on my face as I wondered if the author was high. There is currently a class action lawsuit against Seagate for unprecedented failure rates on their 2Tb drives. I had two 2 Tb Seagate drives go in their 13th month.. both of them died just out of warranty. As I write this I’m trying to get my data off of one of them after getting “cyclic redundancy errors”. Happened a year ago, but now I need stuff on it and hooked it back up. Here’s a link to the class action article.. it’s on a site called “petapixel” in case they don’t let me link….

        http://petapixel.com/2016/02/02/photographers-beware-seagate-slapped-with-a-class-action-suit-over-bad-hard-drives/

  • Jim Dowling

    Great stuff. Do you have any stats for drives’ Bit Error Rates (BER) / maximum unrecoverable read errors (URE) / non-recoverable read error rates ? By my understanding, manufacturer quoted BERs for commodity drives, often 10^14, tend to be 10^15 or higher in practice.

    • Andy Klein

      The SMART stats we find useful are:

      SMART 5 – Reallocated_Sector_Count.
      SMART 187 – Reported_Uncorrectable_Errors.
      SMART 188 – Command_Timeout.
      SMART 197 – Current_Pending_Sector_Count.
      SMART 198 – Offline_Uncorrectable
      We talk about them more here: https://www.backblaze.com/blog/hard-drive-smart-stats

      • mickrussom

        I use 187 – anything on 187 = pull and RMA. Which one do you “trigger” on?

  • Jon

    I’m a bit confused. the Seagate 3TB drives had a horrible failure rate last year. Does the grapho of the cumulative rate by manufacturer include those or it it just drives still in operation? Is it just the sheer number of Seagate 4TB drives drowning out that signal?

    I think at some point you should do an over view of all your data through the years. To many people just look at the latest post and assume that snapshot tells the whole story. In that regard I also miss seeing the chart of drive death by installed age.

    • Andy Klein

      The cumulative graph is includes Seagate 3TB drives as data through when they were removed. So they still have an impact on the over all Seagate number, but you are correct that the shear number of Seagate 4TB drives creates a large impact.
      Your second point is valid as we didn’t include an obvious reference back to previous studies. You can start with the one from the previous quarter: https://www.backblaze.com/blog/hard-drive-reliability-q3-2015. If you want more, just use the search box at the top of the blog and enter “hard drive stats” to get the rest.

  • Randy Morris

    Toshiba is my go to platter drive now. Quiet, fast, reliable, cheap. Just lost 3TB from a HGST, so I’m not so gung ho about reliability there. WD black has served well. Seagate has failed me EVERY time.

    • Pietz

      how can you read this post and base the reliability of a manufacturer on a single drive?

      • Randy Morris

        The HGST was the most current drive, my personal one, that failed. I’ve been a pc tech since ’93, when Seagate, Conner, Quantum, Micropolis and Maxtor were battling it out. I’ve had many a Seagate crater on me, and HGST has had it’s ups and downs, especially with the Deskstar series. I’m praying that my Toshibas are going to last, since they’re HGST made. You assume that I base my statement on just one drive, but trust me, I’ve been through 1000’s professionally, and probably about 60, personally, with family pc upgrades and repairs. Btw, I’m not talking big box store technician, I’ve been a beige box builder/tech for 20+ years.

        • Stephen Wiebelhaus

          I’ve been building workstations for me and family for 20+ years. My best advice, multiple copies on separate drives (aka backup). Anything important to you should be on multiple drives ASAP, even if that means backing up your documents between two Windows PCs.
          Once a drive is 3 years old, it needs to be in a raid mirror or weekly backup to another NEWER drive. With a pair of 5 year old drives, when one fails it’s not surprising to see it’s partner die while copying stuff off to a new one. So have it’s pair or copy be on a newer drive.

        • Stephen Wiebelhaus

          I’ve had three Seagates die in the past 4 years, all were between 1-4 years old at time of failure. Yet, I still have two older Seagates in service, 5-8 years old. My other drives, mostly WD 4-10 years old, and one Toshiba 6 years old. Most recent purchase, HGST almost a year old, was bought due to frustration with Seagates failing, and seeing this report from BackBlaze.

          It may be that around 5 years ago, Seagate just had some quality or process changes that resulted in more frequent drive failures, or maybe they are just more prone to sudden electronics failure in my box, don’t know for sure. But, for me, I’m not running enough drives for the lower price to overcome the higher failure rate, though that works in the BackBlaze use case.

  • genar0

    Thank you for sharing. Here in our province, they only sell Seagate or WD HDs. I guess I have to go for Seagate now.

  • Tech1ceTech2ce

    What is the cost difference between the HGST and Seagate Drives?

    • Jesper Monsted

      Roughly 10% in normal distribution. They may not be able to buy the HGSTs in bulk, though. He mentions that buying 50 of something at a time is just too annoying when seagate will send you 5000.

      • Tech1ceTech2ce

        Sorry should have clarified. I was referring to the new 4TB HGST drives vs the 4tb. Seagate drives that they are primarily purchasing now.

  • JY

    Oddly enough, I’ve been using WD drives for years and never had trouble with them. I generally had them installed in daily gaming desktops as well as linux servers I put together myself. I’ve also used HGST (Previously Hitachi) drives for years. I actually have a Hitachi and WD 1 TB drives are still going strong. I picked up a new WD 3 GB drive not long ago.

    • HenkPoley

      Well, gaming systems often have a short work-life (3 years?). The formula for the chance to have them all working is (1-0.03)^(total years of runtime).

      So say with 2 gaming stations and a Linux server (2+1=3 PCs) for 3 years, it would be (1-0.03)^(3*3) ~= 76% chance that none of them have failed.

  • linkdude31

    No WD Enterprise (RD or SE) or Black drives tested…what?

    • solomonshv

      they don’t use enterprise drives. those cost too much. it’s cheaper to keep replacing consumer drives that keep dying.

  • SomeGuy2398067

    Can you guys explain why you have so few of the 4TB, 5TB and 6TB Toshiba drives? They are quality drives from what I hear and based on my own research, they are priced on par with the Seagate drives. Many people like me use these stats to make decisions on which drives to buy and I was hoping you would be using more Toshiba drives.

    • Andy Klein

      Good question. It has to do with what is available for us to buy at what price. We purchase our drives from the reseller channel. We ask for quotes on say 5,000, 4TB drives, to be available on “X date”. To date, Toshiba resellers have not met those criteria. We have been offered lower (much lower) quantities of Toshiba drives at a higher unit price but that doesn’t make much business sense. We have nothing against Toshiba drives, we like to have multiple vendors, but we just have not been able to get the quantity and price we need.

  • Mark Scudder

    Andy,

    Thank you for sharing this information as you always do. However, I hate to say it, but it’s not enough. You or somebody else at your company has to focus your efforts elsewhere.

    I am a former Backblaze customer. I was almost entirely happy with your service, but I had to get rid of you and go to CrashPlan, which is a wretched company with terrible software and people who don’t care.

    Why did I go from being a happy Backblaze customer to an unhappy CrashPlan user? Long story short, I had to put a Linux server in my basement. I’m actually pretty poor, not like you guys in San Fransisco where even the homeless get $500 on an ATM card every month. So sometimes if I need something badly enough, I have to assemble it from junk parts and throw Linux on it so it’s at least party reliable. Your software supports Mac and Windows, and does so very well, as I recall. But the information on my Linux server is at least as important as the information on my other computers. I’m sure you’ve heard the old advice that your data should exist in at least three places. As badly as it is implemented, CrashPlan does this.

    The other reason I had to switch is because you couldn’t adapt your online backup software to do the same thing, but to my local computer. Part of the role of that Linux server in my basement is so CrashPlan can back up my computer and my wife’s computer locally, so I can get files back quickly in the event of a system failure, or accidental deletion (we both work with big files, and I’d rather spend $140 on an extra hard drive for my server that $140 every time you need to ship me a hard drive; again, the assumption people have the money to just do that is insulting). I also didn’t think it was necessary, since you write such nice software, to burden is with that web interface, which is good for restoring one or two files quickly, but not a large number in an unattended manner, say for instance my house burns down, at which point paying you for hard drives seems a bad financial idea. As bad a company as CrashPlan is, their software supports all major OSs reliably, and if my internet can handle it, I can restore gigabytes of files from their software and just let it run – no maximum ZIP file sizes, no manually moving files into place. Windows and Linux don’t have “Time Machine,” and it’s short-sighted to assume all your users can afford Macs and “Time Capsules” for backup over wi-fi.

    And the other reason: CrashPlan has a “family plan” for $13 and change that allows me to back up 10+ computers, and if I recall correctly, with you it was $5 per machine per month, no alternatives. As clunky as CrashPlan’s software is, I can also protect my dad and my wife’s parents – in essence protecting myself from more free tech support visits – from my CrashPlan account for free. (They also backup to my Linux server, another thing CrashPlan does.)

    You’ll probably likely just ignore this email, or worse, reply and say that’s not your business focus. Well it should be somebody’s. CrashPlan is the only backup service provider who does all this in one relatively easy to use app. But they don’t have any competition. You know their app is written in freaking Java? It runs like crap and often resends terabytes of files it should already have to its cloud servers, tying up my internet for weeks. They don’t have a local deduplication system like you do, meaning if I move a couple hundred gigs of files to another machine, CrashPlan resends them all (and, I should add, “deduplicates” them on the remote end and calls that a “feature”)

    Please. I’m begging you. Now that you are managing your failure rate, please give CrashPlan some competition, so I can come back and be happy and secure with my backups. Maybe you guys are into monopolies out there, but it hurts the little guy. Before the little guy has no voice left, will you hear me?

    • Andy Klein

      Thanks for the post. I’m glad you found a backup solution that fits your needs. Many people find our solution straight forward and easy to use and are happy to pay us for that.
      Let me correct one inaccuracy in your description. If you have a large amount of data to restore, you can order a hard drive from us with your data on it for $189. If you return the hard drive to us within 30 days, we will refund your $189. We do this as many people have monthly bandwidth caps (or restricted bandwidth) and downloading even a couple-a-hundred GBs of data or more is not realistic for them.

    • Matthew

      Amazing comment! I too am in the same boat. Backblaze needs to get on it and develop a native linux client. CrashPlan’s java app is a pain but gets the job done and can do so with my own encryption key. Come on Backblaze, it’s 2016 now; support all the OSes!

      • jameskatt

        Every backup service needs to have a business plan that makes a profit to stay in business.

        Linux has numerous versions. The only way to cover Linux is to use Java. But then you run into performance problems and upgrade problems that CrashPlan terribly has, that Backblaze does not. Since the vast vast majority of consumers use either Windows or Mac, Backblaze probably has to focus its resources on customers that will make it the most profit and keep it a viable company. The few Linux users simply cannot carry their weight in supporting. Backblaze for all the servicing and software upgrades it will need. That would be like asking a developer to support Blackberry when smartphone users primarily use iOS or Android.

        I chose CrashPlan over Backblaze for one feature that was crucial to me: Forever backups of external drives. Since I run on a MacBoo Pro that I take from business site to business site, eventually Backblaze would erase data from an external drive that isn’t connected within one month. CrashPlan on the other hand kept all my data I had on external hard drives I haven’t connected in over a year.

        But despite CrashPlan’s utility, it is actually only useful for computers that stay in one place. Since I am mobile, CrashPlan tends to endlessly scan for changes everytime I reconnect drives. This is why a year later, my MacBook Pro and most used drives have only been 50% backed up. CrashPlan doesn’t have a full backup yet, which is frustrating to say the least. By now, Backblaze would have a backup since it runs natively and rapidly on Macs and Windows.

        I kept CrashPlan for my non-mobile computers and mobile computers (though the mobile ones never finish though I keep hoping). To insure I have an actual backup for my mobile computers and their external drives that are stored forever connected or not, I also simultaneously have been the native app Arq to Amazon’s cloud. Arq completed a backup faster than CrashPlan. So I run both – plus my most used documents on Microsoft’s OneDrive for Business using oDrive.

        Having 3 cloud backup services plus multiple local and remote backups and 3 cloned working computers should keep me satisfied with the security of my data.

        • A native Linux client written in C++ would run on nearly any Linux distribution; if they provided the source code (which I don’t think is impossible, seeing as the API is public anyway), it could even be compiled for ARM. For the UI, QT5 works well and would also work on most distributions.

          • Stephen Wiebelhaus

            Even if they don’t provide source code, a Debian based package would cover a
            large portion of home Linux users and corporate workstations.Many Linux distributions are based on Debian or Ubuntu which is also based on Debian. The users of other distributions are either corporate or more technically inclined and would be more likely to use Wine or virtual machines to run the BackBlaze client, or setup other backup solutions.

      • If you want to backup your Linux OS, check out our new service Backblaze B2 Cloud Storage: https://www.backblaze.com/b2

        This provides storage for must $0.005/GB/month and is accessible via a web GUI, CLI, and API. There are already a couple backup services that support this on Linux: https://www.backblaze.com/b2/docs/integrations.html

        • Oliver Erlewein

          And in conjunction with B2 I suggest using hashbackup.com
          Best frigging way to backup I’ve ever seen!

          • kefoster

            Haven’t heard about this one! Need to give it a try.

    • mzs_47

      We can afford the time machine and time capsule, provided you have
      sometime to use FreeBSD with ZFS, which allows me to send and receive
      the filesystems over SSH.

      If you don’t have much time then, checkout PC-BSD which is trying to make this easy by bringing a GUI to manage this.

    • Terry Fundak

      I’m not sure why you rate CrashPlan so low. I have used and hosted a CrashPlan server for about 5 years and I rate CrashPlan tech support as by far the best of any company I have ever dealt with. Perhaps I am lucky to be a CrashPlan Server owner. Perhaps you are paying the price of not paying the price of business quality backup. There software is rock solid and when I do have an issue, I have them on the phone in minute and I allows get to the end of the tech support session with my issues handled at an expert level. In my mind, tech support is the only thing the makes or break a product in the long term. No company can do world class tech support for free. Engineers are way to expense.

  • pooloo

    How are these stats pertinent to a typical consumer? Data centers typically have their drives running 99.99% of the time, while the average consumer would not, which is why there is a distinction between consumer and enterprise product. How is this a fair depiction of consumer based product reliability? Shouldn’t you be using drives designed for these particular services, would that not save you money in the long term?

    • Andy Klein

      In our experience there is little difference between Enterprise and Consumer drives for our purpose. In truth all drives fail and we’ve designed our own hardware and software in recognition of this reality – it doesn’t care if the drive costs $150 or $1,500.
      People can decide if this information in valuable to them. They are free to add it other sources of information such as personal experience with hard drives, drive tear-downs and reviews, and manufacturer specs and marketing.

    • Ray Tuholski

      4 9’s is not terribly reliable for enterprise needs…. that would get you fired from certain IT shops. If you read through the BackBlaze blog, they lay out their philosophy regarding their server builds pretty clearly. That philosophy says “Why buy a $400 drive to do the same work that a $100 drive can do?” Nothing like watching your bonus vanish and your CAPEX budget go down the drain buying a bunch of unneeded, expensive hard drives.

    • Hellscreamgold

      nope

    • David Antin

      For the consumer the point is that Backblaze data suggests that RAID using an array of inexpensive consumer drives produces a better ROI than using enterprise class drives.

  • noyb

    WD really surprises me, their platter drives were always my goto for reliability but that is truly abysmal reliability for SSD

    • Lastb0isct

      Where are you seeing SSD listed? I’m pretty sure these are all HDD, platter drives.

      • jameskatt

        I suspect SSDs would be killed in the heat of Backblaze’s pods. Data on SSDs is not as reliable in heat as with hard drives.

        • Xyriin

          A large part of that heat is due to the mechanical drives. With SSDs you’d find they operate cooler, use less power so less ambient heat, and they need less operation time as well for more idle time leading to cooler temps.

  • Can you show us photo of the server – how HDDs are mounted to the rack?

  • Bill McKenzie

    Problem with the 2tb A330 HGST drives on eBay and Amazon is they are all (AFAIK) used drives with thousands of hours on them already. I went through 2 and both died within a month.

    • Andy Klein

      HGST has been moving towards the enterprise space, so they haven’t made the type of drives we use in a while. Sometimes we can still buy some new ones in the channel, but they are hard to find.
      Appreciate you letting folks know about Used Drives, folks should pay attention to that when they buy. We don’t buy used drives, so I can’t shed any light on the reliability of used drives.

      • The worst thing about Amazon is that it’s very hard to even know you’re buying a used drive — sometimes I think that as long as Amazon gets their cut from the seller, they’d gladly just help sell you a hard-drive shaped enclosure with a bunch of marbles inside.

  • Definitely going to look into some of these HGST drives. Seems like extremely low failure rate and lots of great stats.

  • Dc120

    Thank you so much for sharing this data!

  • Andy Klein

    Here’s a link to the 4TB Seagates drives on Amazon: http://www.amazon.com/Seagate-SATA-3-5-Inch-Desktop-ST4000DM000/dp/B00B99JU4S

    I agree some of the older HGST ones in particular are hard to find.

    • Emmanuel Goldstein

      Even when the original Q4 post came out, trying to find any of the HGST models was a pain since they’ve all been EOL’d for a while.

      • HenkPoley

        Yeah, it’s up to you wether you want to pay the 57% markup for the added reliability of HGST.

    • Well, if the link to AWS is accurate, these Seagate 4TB HDDs spin at 5900 RPM, which may make them more reliable over time. Wear and tear matters and a HDD that spins at a lower number or RPMs may enjoy a longer life. And in this particular application, performance is probably not as important as reliability. And of course, the price is right.

  • Emmanuel Goldstein

    Always fun to look at the stats, but it’s always from drives that you can’t find anywhere.

    • mickrussom

      Agree. It would be nice if they bought 200 drives of various flavors that are available at a good price per TB at cdw/newegg/amazon/insight/etc. I think this points to HGST being the best, but their prices are too high according to them so they buy the higher failure rate drives. Interesting.

      • Joe Chahine

        Yea they should buy drives release in 2016 is give us stats on their 3 years reliability.

    • Milk Manson

      Agree. They need to give us long term reliability stats on current model drives.