Backblaze’s Q1 2018 HDD reliablity report shows record-low failure rates

Turning the calendar from April's picture of a tabby kitten with a yarn ball to May's photo of two calico kittens curled up inside coffee mugs means we are one-third of the way through 2018. It also means that Backblaze is back with its latest hard drive reliability report. The cloud backup outfit provides data about the reliability of drives in its enormous storage pool every three months, and the results this quarter are the best so far.

At the end of the quarter, Backblaze pooled together reliability data from 98,046 drives that met the criteria for inclusion. Long-time readers will know that Backblaze tends to use more drives from Seagate than from other manufacturers, and that trend continues this quarter—just four Seagate models represent over 73% of all drives in the report. Furthermore, all of Backblaze's drives with capacities larger than 8 TB come from manufacturer. All told, the company tracked data from fifteen models from Hitachi, Seagate, Toshiba, and Western Digital.

The annualized failure rate for Q1 2018 fell to just 1.2%, down from 1.65% in the previous quarter. Seagate's 4-TB ST4000DM000 had the highest single-model annualized failure rate at 2.3%, with WD's 3 TB WD30EFRX following closely at 2.25%. Those two models are some of the smallest-capacity models that Backblaze keeps in operation, suggesting they may also be among the oldest.

When looking at reliability over the hard drives' lifetime, the annualized failure rate across all manufacturers rises to 1.84%. The big losers in this metric are a pair of WD drives, the 3-TB WD30EFRX mentioned above and the company's 6 TB WD60EFRX model. We must note that Backblaze ran fewer than 450 of each of those models, a small sample size compared to the pool of nearly 31,000 Seagate 4 TB drives. That 1.84% annualized failure rate is the lowest the company has ever seen, beating the previous best of 2% in the previous quarter.

Backblaze has found exceptionally low failure rates for Toshiba drives over the last couple of quarters and is introducing 8-TB models from the manufacturer into its pool. The 20 units in use aren't enough for inclusion in the company's reliability report, but the company did point out that those drives have been flawless so far. Backblaze also notes the number of helium-filled drives from Hitachi and Seagate is growing and teased comparisons of helium-filled and standard drives in the future.

Gerbils that want to take a closer look at the company's figure can download Backblaze's data in raw form here or read the Q1 2018 writeup by following this link. Those that choose to pore over the numbers themselves might notice that the company has started recording five new SMART attributes, all of which are related to SSDs. Backblaze experimented with speeding boot times of its servers with Samsung 850 EVO SSDs before deciding the improvements weren't worth the extra cost.

Comments closed
    • Growler
    • 1 year ago

    Backblaze’s back? Alright!

      • Redocbew
      • 1 year ago

      Why would you do that?

    • Neo1221
    • 1 year ago

    [quote<]The big losers in this metric are a pair of WD drives, the 3-TB WD30EFRX mentioned above and the company's 6 TB WD60EFRX model. We must note that Backblaze ran fewer than 150 of each of those models...[/quote<] Am I just reading the stats wrong, or does it say there were 437 WD60EFRX drives and 180 WD30EFRX?

      • WayneManion
      • 1 year ago

      A slip of the finger. We have corrected the article to say fewer than [i<]450[/i<] of each model.

    • ronch
    • 1 year ago

    Thinking of getting a 2TB or 3TB drive. Go with Seagate?

      • DancinJack
      • 1 year ago

      Not gonna matter that much I’d say. Though I would advise you to just go 4TB or larger. You may not NEED the space right now, but IMO 4TB starts the sweet spot for HDDs these days.

      • jihadjoe
      • 1 year ago

      IIRC the Seagate 3TBs (especially the DM001) had a pretty bad fail rate. It’s with the 4TB and up that they recovered. I’d stay away from the 3TB Seagate because there’s no way to guarantee you won’t get sent a DM001.

      It’s usually just one or two specific products that make companies look bad on reliability surveys like these, just like WD with the 2TB EARS/EADS Greens. Even HGST had the infamous Deathstars back when they were still IBM.

    • highlandr
    • 1 year ago

    Wait, what kind of calendar do YOU have?

      • Vinceant
      • 1 year ago

      Seems to be full of pussy.

    • DancinJack
    • 1 year ago

    Seems most everyone is doing better, but I will stick to my HGST models until something really forces me to change.

      • techguy
      • 1 year ago

      HGST and Toshiba at the top again. The only 2 brands I buy anymore.

        • Waco
        • 1 year ago

        You can keep buying them, yes. Statistically it won’t matter which you get for a small population (or a large one, really).

    • Waco
    • 1 year ago

    I wonder how many of these reports it will take to finally remove the tarnish they applied to Seagate a few years back. I’ve had great luck with all manufacturers in the past 5 years in terms of AFR.

      • willmore
      • 1 year ago

      I swapped out the oldest drives of my RAID a few months back. That left me with a stack of drives that needed testing to see if they might be useful for other things. Once I sorted through those I had some old (50-70K hour) Seagates, one HSST, and several dead WD. No live WD at all. No dead drives from any other brand. *one* Seagate with 72K hours had some reallocated sectors and had a low value for ‘remaining lifetime’ due to the 72K hours. All of the rest of the metrics were good.

      These were all 2T drives, but they’re from several different product lines of each manufacturer–except HGST, both drives of theirs I have are the same model.

      Not backblaze level testing, but their drives undergo a different selection bias, so I wouldn’t expect any correlation.

      • Captain Ned
      • 1 year ago

      The 7200.11 debacle did for Seagate what shoddy RAM did for OCz. They both got better (they’re no longer newts), but the stain continues across all future product lines.

        • Spunjji
        • 1 year ago

        I got burned (not personally) by their original market-first 500GB 7200rpm 2.5″ HDDs. I recommended a notebook with that in to my sister… it failed 3 times under the notebook’s warranty before that expired, then it failed again for good luck.

      • EzioAs
      • 1 year ago

      As Captain Ned said, it was necessary for people to know how bad those drives were and Seagate needs to earn back people’s trust.

        • Waco
        • 1 year ago

        It was literally a decade ago. :shrug:

          • EzioAs
          • 1 year ago

          I was actually referring to the [url=https://en.wikipedia.org/wiki/ST3000DM001<]ST3000DM001[/url<], which was first available on 2011. My mistake if the drive Ned was referring to was a different model. However, if there were 2 different models at 2 separate times that have high failure rates, my original point still stands.

            • Waco
            • 1 year ago

            Yeah, that one was more recent. They’re doing well since then though.

      • Chrispy_
      • 1 year ago

      I’ve been buying IronWolf Pro disks for a while now (both home and enterprise archive/cold).

      WD seem to have overcomplicated their lineup by adding too many models and pricing the decent drives (WD Gold, WD Red Pro) with this false premium.

      Disk failure rates are so low now that even when I’m buying 25 disks at a time I’m expecting to never see an issue on any of them in the 3 year service life most of them are subjected to. Since reliability is no longer the primary worry, my attention has turned to artificial market segmentation and removal of useless features for the sake of widening the product portfolio. All I ask is that a disk be free of shingles, have vibration sensors, and a decent warranty.

Pin It on Pinterest

Share This