Backblaze announces FY 2021 HDD drive stat Failure Rates Data
Click here to post a comment for Backblaze announces FY 2021 HDD drive stat Failure Rates Data on our message forum
alanm
Seagate proves once again that it is the least reliable HD brand. As has been the case for last 15-20 years. The higher the capacity, the worse the numbers are. Perfect timing for potential consumer, just after SG announced their new 22tb drive.
anticupidon
As expected, HGST and Toshiba have the lowest failures.
I know, i know is WD after all, but still.
Öhr
Still, one shouldn't apply the average to EVERY product of a company. E.g. the WD RE3 1TB were terrible. Believe me, I had 3 of them: SMART was not permanent (every power off resetted most values, such as power on time), they produced a lot of heat, hated heat at the same time and compared to my other drives (e.g. 5x Seagate Archive 8TB SMR), the failure rate is 2 out of 3 (vs Seagate 0 out of 5). not to mention the WD Red SMR bullshit they pulled just recently...
And same for Seagate: while that one 14TB model does look terrible, their 6TB model still produces the lowest failure rates out of all the drives tested.
Every company has at least one dud each generation. just make sure to avoid it and don't just blindly trust a brand.
mackintosh
Agreed. I can't think of any HDD manufacturer with a spotless record. I've had every major die on me, going back to 1991.
DeskStar
Ah yes Seagate. The manufacturer I'll never buy from again.
At LEAST six HDD's have died on me since 2016. All Seagate! Most from one batch it would seem.
Just sad man. I've had some seriously irreplaceable shite on those drives even backed up to another one and they FAILED! Days before a DVD DL burn of what was needed.
Just have to rely on the old mind box to keep these ones (memories) around for a while longer...
rl66
Same in the company with the 1st WD helium 8To, hopefully in high secured RAID, the fail rate conduct us to chose Seagate to replace them (i was very worried about it).
So yes they do a lot of spindy noise ( Tzzzzzzziiiiiiiiii clock twiiiiiiiiiiz ) and more heat, but they never fail (replaced only due to end of life) the experience is successfull.
Each brand have a black sheep that can discredit the innovation and work, remember IBM disk...
Also always wait the feedback of a product BEFORE to buy it for yourself, let early adopter tell you if good or not lol.
rl66
Never ever buy 6 HD from the same batch !!!
It limit the fail rate.
fry178
Not sure how much transfer of tech between the brands, but once seagate acquired maxtor,
their drives seems to have taken a dive too.
and sure, all brands have failures, but any "caution" drive from toshiba/wd i was able to copy user data from,
when seagate wouldnt, and in overall, i personally had more of my seagate drives go bad,
than all brands combined used by family/friends.
SHS
I can say the same about all other drive too and been using Seagate Drive for all must long they have been around high school 1983 with IBM XT with ST-506
Well I had WD fail are just bad
Astyanax
Consider the failures in ratio to their usage in this data, they are the most used drives by backblaze so of course they'll have a higher failure level.
alanm
Absolutely. Which is why they use AFR (annualized failure rate) as a measure. A Toshiba 4tb has only 2 drives fail out of 97, but shows a worse AFR than a Seagate 8tb with 214 drive fails out of 14,334. Toshiba AFR = 2.04% vs Seagate AFR = 1.49%.
This whole study would be pointless if it was just no. of drive fails irrespective of other considerations.
Venix
@alanm sure but just 97 drives are not big enough sample it can be just luck too.
alanm
Yes, I think minimum sampling should be 1000 drives for statistics to make sense. But Blackblaze is a storage and backup company, they are simply reporting all the drives they have in use, not conducting a scientific analysis to determine best reliability. We choose to make what we will of the numbers.
Venix
I would say 10k but I would have been happy with 1000 from each, and yeah actually i am happy they do ! Such test is not economically viable, what would have been awesome is if every cloud service and back up company would do the same report ! Most likely would have been enough data for almost every drive !
DeskStar
Well I know I'm not investing in HDD's as it is. All nine drives all SSD today. M.2/2.5" to AIC's raided.
I mean yeah I've had SSD's fail on me, but that was back in the day of their inception. OCZ baby!!
112GB SSD cost you $550, so naturally I had to buy two and raid them even though raid wasn't truly a supported thing then for the drives. Mostly due to garbage collection on them.
Venix
I was waiting for ssds to get about 1 euro per gb ratio and got an 120gb Intel 330. Was 2012 and all my friends were telling me "you do not need ssd it's a waste of money" then they saw it in action and literally everyone got one with in few months!
Oh the raid with out garbage collection and trim must have been brutal on em .
fry178
I remember few years back it started to show most ssds die on the controller (about 4 times more likely) than the nand.
Guess my old Vertex heard the call and decided to call out sick when i tried to copy files 😀
Was used as my friends 2nd (external) backup, so at least no recovery needed..
funny how a Vector 180 can still keep up with M2 when it comes to rnd 4k
DeskStar
It made sure to age them prematurely that's for sure...
$500+ a piece at that time I mean....yeah.... Warranty took effect the one time because it was sooner than later when it happened.
