As someone who runs a tiny archiving operation (about 70 hard drives) I find these stats invaluable and am so grateful to you for publishing them! You make it a lot easier to navigate an expensive and intimidating world of hardware options.
I've enjoyed the Backblaze hard drive failure rate posts for years, but never realized they publish the underlying data! <a href="https://www.backblaze.com/b2/hard-drive-test-data.html" rel="nofollow">https://www.backblaze.com/b2/hard-drive-test-data.html</a> (scroll down)<p>It looks quite detailed--a daily snapshot of all of the SMART values for each individual hard drive.
How have they got 2k boot drives for 100k data drives? Are these special motherboards with 50 (!) SATA connectors on them? Or is it many machines netbooting and share-mounting root from a handful of control nodes? I get that it's backup, so most of the data is at rest most of the time, but that still seems like a really high ratio of network ports to disk drives.<p>Either way, interesting that the boot drives are actual drives and not SD card pairs or something, like how vSANs are often set up.
I'm really interested to see how those high capacity Helium-filled drives hold up over time. (i.e. I've been wondering how much leakage is going to be a problem)
Does Backblaze have any SMR drives in use, or do you have plans to start using them? Would be really interesting to see if these exhibit a worse failure rate.<p>Anyway, thanks for sharing this data, I really like the openness.
I thoroughly enjoy reading these stats every time they're released.<p>It's just a shame that their user interface is missing many features I want, otherwise I would use them.