Backblaze has released the 2018 installment of its Hard Drive Stats report. These reference materials reveal how many drives the cloud storage provider is using, how often those drives fail, and how those failure rates compare to previous years. They can also point out broader trends in the storage industry--such as the increased availability of higher capacity drives--for anyone who doesn't usually pay attention to where things are heading.
The report looks promising: annualized failure rates across every category of HDD is on a steady decline. Backblaze said that 1.95 percent of its drives failed in 2016, 1.77 percent failed in 2017, and 1.25 percent failed in 2018. The absolute numbers of failed drives rose year-over-year because the company is using more drives than before, but in general, it seems that more of the company's HDDs are surviving their daily hardships than ever.
Those figures are based on the company's usage of 104,778 hard drives spread across 15 models made by HGST, Seagate, Toshiba, and Western Digital. (It also has another 176 drives excluded from testing because it doesn't use at least 45 of the same model, as well as 1,965 boot drives.) You can see a full breakdown of how often a particular model has, well, broken down over the last three years in Backblaze's comparison chart:
There's a marked difference in failure rates between the new 12TB and 14TB drives the company started using this year and its other HDDs. This could be caused by problems with a specific product; it could also mean that Backblaze hasn't yet figured out how to properly integrate these drives. Either way, there's a big jump from the 0 percent failure rate of Toshiba's 5TB MD04ABA500V and that same manufacturer's 14TB MG07ACA14TA.
A caveat: Tom's Hardware can't speak to the data presented by Backblaze. The company is using far more drives than we could ever reasonably test, and even if we did want to blow the budget for the next decade on a bunch of spinning discs, we wouldn't use the exact same configuration as Backblaze. We're presenting the information as-is and hope it's clear that we aren't advocating for or against certain products based on this report.
And, to be fair, Backblaze makes similar caveats in the report itself. In the preface for this year's installment, the company acknowledged that it called the first Hard Drive Stats report published in 2014 “What Hard Drive Should I Buy,” and said, "that might have been a bit of an overreach" in hindsight. Notice that it's no longer recommending drives to consumers; it's merely sharing information about the performance of the drives its business relies on.
The information behind Backblaze's report is available on the company's website. It's also planning a "Backblaze Hard Drive Stats for 2018" webinar for January 24 at 10am PT for subscribers to the Backblaze BrightTALK channel. This webinar is expected to "dig deeper into the quarterly, yearly, and lifetime hard drive stats and include the annual and lifetime stats by drive size and manufacturer."
