• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

SSD failure rates

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Mr Alpha

Senior Member
Joined
Jul 15, 2006
In my SSD research I stumbled over a french site that had gotten its hand on the the failure rate for different hardware according to a French etailer. Here are SSD failure rates according to brand:
  • Intel 0,59%
  • Corsair 2,17%
  • Crucial 2,25%
  • Kingston 2,39%
  • OCZ 2,93%
No surprises there. Intel is the clear leader and OCZ is worst. As a comparison here are the failure rates of some hard drives:
1TB drives:
  • 5,76% : Hitachi Deskstar 7K1000.B
  • 5,20% : Hitachi Deskstar 7K1000.C
  • 3,68% : Seagate Barracuda 7200.11
  • 3,37% : Samsung SpinPoint F1
  • 2,51% : Seagate Barracuda 7200.12
  • 2,37% : WD Caviar Green WD10EARS
  • 2,10% : Seagate Barracuda LP
  • 1,57% : Samsung SpinPoint F3
  • 1,55% : WD Caviar Green WD10EADS
  • 1,35% : WD Caviar Black WD1001FALS
  • 1,24% : Maxtor DiamondMax 23
Full disclosure: I have a Hitach Deskstar 7k1000.B. Maybe I should be worried.
2TB drives:
  • 9,71% : WD Caviar Black WD2001FASS
  • 6,87% : Hitachi Deskstar 7K2000
  • 4,83% : WD Caviar Green WD20EARS
  • 4,35% : Seagate Barracuda LP
  • 4,17% : Samsung EcoGreen F3
  • 2,90% : WD Caviar Green WD20EADS
As an aside, turns out Steve Gibson is right. Failure rates seem to scale linearly with size.

EDIT: A quick Google translate gives this additional information. This is failure rates for the first year only, ie failures within a year of the sale. So it is only infant mortality rates and so does not include any wear-out-the-flash failures. Also, it only includes the drives that have been returned to the store, not drives RMA'ed directly to the manufacturer, which mean manufacturers with good customer RMA service probably have their failure rates underestimated.
 
Last edited:
Awww crap....

I have two 1TB SeaGate Barracuda 7200.12 in RAID 0 my tower and two 2TB Seagate Barracuda LP in an external RAID 0.

Funny thing is people give seagates a bad rap but I have been running my 7200.12 for a long long time with zero issues. Also no issues with the LP's and I've been hammering them.

Also have a 100GB OCZ Vertex 2 SSD which I bricked my self on a failed firmware upgrade and had to RMA, luckily OCZ's customer service is outstanding.
 
The failure rates on the SSD's are towards the better end of 1TB mechanical disks, Intel is actually astounding!

What I mean is that these stats have been going around, and everybody concentrates on how good Intels reliability is (and it is amazing), and how bad the other SSD manufacturers are; but they are still doing pretty well comparing to mechanical disk manufacturers... And who has been in storage for what seems to be ages? The "other" SSD manufacturers are all "ram companies" who have probably been used to RMA's being ok, hopefully stats such as these will cause them to up their game.
 
interesting bit of information, thank you for that!

and yeah, nearly 10% failure rate in the first year is higher than I expect them to be... it's almost like gambling your data =| unless you raid or regularly back up of course
 
interesting bit of information, thank you for that!

and yeah, nearly 10% failure rate in the first year is higher than I expect them to be... it's almost like gambling your data =| unless you raid or regularly back up of course

Raid is not a substitute for backups...

And since this is a tech site I would assume everybody with any important data does back it up, though maybe it does require being burnt once before you understand how important it can be. :)

I think disk failures are still quite a big hassle backups or not.
 
Wish we could see the sample size :(
Your not the only one. Would be nice to have some error bars on that. In the beginning of the article it mentions that statistics are only given for brands when there is more than 500 sold and for models when there is more than 100 sold. This does mean you can do some guesstimates. A bit of napkin math gives me (for a 95% confidence interval):
  • Intel (-0.0813%, 1.26%)
  • Corsair (0.893%, 3.45%)
  • Crucial (0.950%, 3.55%)
  • Kingston (1.05%, 3.73%)
  • OCZ (1.45%, 4.41%)
This is on the pessimistic assumption they all sold only 500. With the real data the error bars should probably be smaller.
 
Your not the only one. Would be nice to have some error bars on that. In the beginning of the article it mentions that statistics are only given for brands when there is more than 500 sold and for models when there is more than 100 sold. This does mean you can do some guesstimates. A bit of napkin math gives me (for a 95% confidence interval):
  • Intel (-0.0813%, 1.26%)
  • Corsair (0.893%, 3.45%)
  • Crucial (0.950%, 3.55%)
  • Kingston (1.05%, 3.73%)
  • OCZ (1.45%, 4.41%)
This is on the pessimistic assumption they all sold only 500. With the real data the error bars should probably be smaller.

Yeah I ran a quick calculation based on the same numbers.

I guess the time I spent in school and internship using statistics for homicide research with SPSS and Access has made me only interested in infallable numbers :) Who would peer review a data set on hard drive failures? :sly:

Not that these aren't interesting numbers, just wish there was more behind it so we could assign higher relevance to the results.
 
I can agree with those numbers.

In my old job we had a good number of black 1TB and black 2TB prematurely die within the year, along with the WD green 2TB (EARS) model.
 
Really nice stuff here. It seems they're planning on running something like this every few months or so. Last one was in April 2010 (Pg 6 for HDs). They got themselves quite a nice source.

Your not the only one. Would be nice to have some error bars on that. In the beginning of the article it mentions that statistics are only given for brands when there is more than 500 sold and for models when there is more than 100 sold. This does mean you can do some guesstimates. A bit of napkin math gives me (for a 95% confidence interval):
  • Intel (-0.0813%, 1.26%)
  • Corsair (0.893%, 3.45%)
  • Crucial (0.950%, 3.55%)
  • Kingston (1.05%, 3.73%)
  • OCZ (1.45%, 4.41%)
This is on the pessimistic assumption they all sold only 500. With the real data the error bars should probably be smaller.
Actually, I'd expect a more complete analysis to have larger error bars, due to clustering, since the data comes from only one store. For instance, the April article has one hard drive (Seagate Barracuda 7200.11 500 Go) with an RMA rate of 39%. I'd be surprised if that's anywhere near the actual failure rate, and is rather an artifact of the restrictive sample cluster. Of course, we can't do a better analysis since we have no idea how clustered these observations are. :(
 
A mate of mine who manages a server room containing over US$1 million of servers said that consumer level drives tend to have too many reliability issues so they have 400GB server drives. Go figure.
 
We've had less issues with SCSI and SAS drives on servers too, though these days it's pretty rare that anything apart from the OS and applications are located on the internal disks.
 
tell you have one go bad on you like i did. had bad sectors.

Yea, I think one of my raptors is going now after a year and a half... getting occasional bsods and when i reboot i get intel matrix raid manager warnings on one of the drive. So far, I've been able to mark the drive as normal and keep my array going... but i doubt that will last.
 
If only Google had given up it's data on platter drive fail rates we would likely not even have to worry about the sample size.
 
Back