Last month, online backup provider Backblaze published some interesting statistics on hard drive failure rates in its "storage pods." Now, the company is back with new data that compares failure rates for consumer- and enterprise-oriented drives. The numbers suggest that consumer models are just as reliable as their server-grade counterparts.
Backblaze houses user data in custom "storage pods" that have logged 14,710 consumer drive years thus far. (One drive year is equivalent to a single drive running for a full year.) There have been 613 failures, which translates to an annual failure rate of 4.2%. On the enterprise side, Backblaze has tallied 368 drive years with 17 failures: a 4.6% failure rate.
Some of those enterprise hours come from an experimental storage pod populated exclusively with server-grade drives. The rest are from off-the-shelf servers that handle "transactional data such as sales records and administrative activities." As Backblaze notes, there are some important differences between the two implementations.
The enterprise drives are used heavily. The consumer drives are in continual use storing users’ updated files and they are up and running all the time, but the usage is lighter. On the other hand, the enterprise drives we have are coddled in well-ventilated low-vibration enclosures, while the consumer drives are in Backblaze Storage Pods, which do have a fair amount of vibration.
The company adds that the failure rate in its experimental enterprise pod is "statistically consistent" with the failure rate of the consumer drives. There's another caveat, too. Although some of Backblaze's consumer drives have been running for as many as four years, the enterprise drives have only been active for two years. The company has observed that failure rates climb steeply after three years, and it isn't sure if enterprise models will exhibit the same uptick.
Despite the imperfect data, the results should be encouraging to folks who want to save a few bucks on mechanical storage. It's also encouraging to see companies sharing data like this, even if they continue to steer clear of naming specific manufacturers and products. Thanks to TR reader Gleb for the tip.
|1. Hdfisise - $600||2. Ryszard - $503||3. Andrew Lauritzen - $502|
|4. the - $306||5. SomeOtherGeek - $300||6. Ryu Connor - $250|
|7. doubtful500 - $200||8. Anonymous Gerbil - $150||9. webkido13 - $135|
|10. cygnus1 - $126|
|Radeon R9 Fury X voltage scaling results are underwhelming||37|
|Google begins removing Google+ integration from its services||7|
|Razer purchases Ouya's software platform and technical team||5|
|95% of Android phones vulnerable to Stagefright MMS exploit||38|
|The TR Podcast 181: In which we avoid talking about Skylake||2|
|Asus' X99-M WS mobo makes for bite-size workstation builds||19|
|Samsung's SE370 FreeSync displays wirelessly charge your phone||16|
|Valve closes Steam password reset security hole||5|