AAPL Stock: 117.81 ( -0.22 )

Printed from

BackBlaze report claims temperature does not affect drive failures

updated 03:52 pm EDT, Mon May 12, 2014

Research on 34,000 hard disks finds no correlation between failure and temperature

The operating temperature of a hard drive does not directly affect the failure rate, according to research by backup cloud service BackBlaze. After analyzing over 34,000 drives, the company found there to be no overall correlation between failure and temperature when looking at the data as a whole, but some drives were found to be affected by heat.

As part of the research, the company discovered the drives do operate at different temperatures, with almost nine degrees centigrade between the coolest and warmest averages per model of drive. BackBlaze typically uses 45 drives in each of its custom Storage Pods in the data center, with the spread of models in individual pods and in varying locations of the data center being "somewhat random," leading engineer Brian Beach to believe the environment is not a contributing factor, and that the temperatures stemmed from the drive's action. The majority of drives were held within the range of 15 degrees centigrade and 30 degrees centigrade, a relatively comfortable working temperature range far below the typical manufacturer maximum of 60 degrees.

While for the most part the drives did not correlate between failure and temperature, there were some drives that seemingly did. One Hitachi Deskstar drive and three Seagate Barracuda drives were considered significant in Beach's mathematics, with the 1.5TB Barracuda and Barracuda LP seemingly being the most significant. While both typically ran cooler than average, the two drives suffered failure rates of 11 percent and 34.6 percent respectively when ran above the average temperature, with respective rates of 7.9 percent and 15.6 percent below.

A graph of failure rates for the Seagate Barracuda 1.5TB drive showed relatively even amounts of failure at 24 degrees and below, with the rate almost doubling to over 13 percent at 25 degrees, rising again to the hotter end of the table.

Beach advises that, aside from the one specific drive, it does not matter whether a drive is slightly warmer or cooler than average, for the most part. "As long as you run drives well within their allowed range of operating temperatures, keeping them cooler doesn't matter."

BackBlaze has performed a considerable amount of research on the large number of desktop drives it uses as part of its service. Previously, it has provided feedback on reliability by manufacturer and average drive lifespans.

by MacNN Staff



  1. DiabloConQueso

    Fresh-Faced Recruit

    Joined: 06-11-08

    Roughly the same thing Google researched and discovered in 2007:

  1. chimaera

    Dedicated MacNNer

    Joined: 04-08-07

    The Google paper is very interesting. But they withheld brand & model data from their listings.

  1. AlenShapiro

    Fresh-Faced Recruit

    Joined: 04-24-00

    Yes But.....

    A typical user does not leave a drive on 24/7 in a temperature controlled environment. A typical user spins the drive up several times a day and sleeps their computer quite often. This behavior is death to drives. It's not the steady state use or spin down to cold that does the real damage, it's the spin up from cool. Although drive manufacturers try to match up thermal expansion characteristics of the metals they use, they tune their alloys for expected steady state operating temperature ranges. If normal users are to make use of this information they must spin down and/or sleep their drives as seldom as possible for it to be relevant.

    It's similar to a light bulb manufacturer declaring that their bulbs last 10,000 hours on average but failing to mention that greater than 500 power cycles from off to on will break the filament.

    Hopefully user-level SSDs will make this point moot but how many people take delivery of a new hard drive based computer or separate hard drive in winter and turn it on from cold (sometimes freezing). Not realizing that they have just severely reduced the life expectancy of their hard drive (hint warm it up to, at least, room temp first).

  1. Makosuke

    Forum Regular

    Joined: 08-06-01

    I really like BackBlaze even if I don't use their service, because they are the *only* organization on the planet giving statistically meaningful, real, useful data on hard drive reliability. For all the pontification on the internet about what brands or models are better or worse, or what actual failure rates are, it's all anecdotal. Google was the one organization that could have given useful data, but they withheld model and manufacturer info, rendering it useless for anything but general idea. StorageReview attempted to compile a database of user reports, but it never had the sample size to be useful, and was biased by self-selection anyway.

    Backblaze, however, that's data you can actually use. Practically worth keeping them in business just for the data!

    @AlenShapiro what you say is entirely true. In theory. The only issue is that without any concrete data on thermal cycles, there's no way to say for sure just how much of an effect a single thermal cycle has on a drive's reliability. To use a (poor) example, a filament light bulb might last 8000 hours if left on continuously, but loses 10 hours of life every time it thermally cycles, so it only ends up lasting 750 hours in practice. A compact fluorescent bulb might last 20,000 hours left on continuously and lose 2 hours every time it's cycled on and off, so it ends up lasting 8000 hours. And an LED might last 20,000 hours continuously and lose 0 hours per thermal cycle, so lasts 20,000 hours no matter what you do to it.

    Point being, the amount of damage a thermal cycle does to the overall lifespan can and probably does vary greatly depending on the technology.

    Lacking any actual statistics (the manufacturers probably know, but aren't saying) on exactly how detrimental to the overall lifespan a thermal cycle is to a hard drive, there's no way to say exactly how much shorter the lifespan of a consumer hard drive is going to be relative to a drive run continuously like this.

    For example, if a drive loses 8 hours every time it's thermally cycled, and is run for 8 hours a day at an office but left off for 16 hours, it's going to last *longer* being turned off and thermally cycled every day than it would have run continuously.

    It might not, but there's no hard data to prove one way or the other.

Login Here

Not a member of the MacNN forums? Register now for free.


Network Headlines

Follow us on Facebook


Most Popular


Recent Reviews

Ultimate Ears Megaboom Bluetooth Speaker

Ultimate Ears (now owned by Logitech) has found great success in the marketplace with its "Boom" series of Bluetooth speakers, a mod ...

Kinivo URBN Premium Bluetooth Headphones

We love music, and we're willing to bet that you do, too. If you're like us, you probably spend a good portion of your time wearing ...

Jamstik+ MIDI Controller

For a long time the MIDI world has been dominated by keyboard-inspired controllers. Times are changing however, and we are slowly star ...


Most Commented