Backblaze is a cloud storage provider and they have been posting stats on hard drive reliabilty. The model specific results are worth a look.
2016
2017 Q1
2017 Q2
2017 Q3
7 Likes
This has been talked about a lot in the past.
My take on it: Take those results with a giant grain of salt, because the conditions they are using those drives under are nowhere near real-world scenarios for the average user.
2 Likes
My problem with backblaze is that they don’t use the same amount of drives from each brand. And that the drives day row is retarded that they combine all of the drives power on time. I didn’t know that seagate 4tb drives have been around for about 8637 years.
It’s kinda logical… how else would you calculate an average…
I might be missing something but it doesn’t look like they are doing that. Look at drive days it makes no sense.
To their credit, you can download the data and do your own analysis.
Haven’t read the latest reports the last few years (because… see above), they used to though.
In a way it makes sense though. Whether they list the average POH or the total POH summed up doesn’t make that big of a difference and neither of them is “accurate”.
It’s just to get a general idea anyway, for deeper analysis you’re better off downloading the statistics and do your own analysis anyway, as Eden mentioned.
But if that’s worst case scenario, then wouldn’t that probably mean it’s the best comparison to make?
Users would choose whichever drive survives the most in the worst case scenario. The odds of that changing drastically in less intense scenarios seems like it’d be low?
The thing is that in the real world most drives will either die in the first 6 months or live forever (where forever is like 10 years, RIP my 10 year old drive).
On the other hand you can get the last surviving drive of Backblaze and still be unlucky and have it die after a year.
When it’s all said and done, it’s a decent indicator of the BackBlaze drives. You notice that this year they have a pretty good bunch. Nothing really stands out unlike prior years. You betcha that the HD manufacturers are taking notice. The nice thing about the drives that they use is that they’re not special. Exact same commercial drives as everyone else and sometimes not even enterprise grade.
That’s the problem though, they are not designed to run in this environment, of course they die faster then for the regular user.
1 Like
Doesn’t matter. As long as you only use the data points from BackBlaze then it’s valid. There is enough info from all the years that they have been doing it to start seeing trends and make analysis. It’s at least interesting and potentially useful. Certainly from the data there I made the decision never to touch a 3tb hd.
1 Like
I would day it’s useful for specific models of drives, but not really for comparing brands. They’re using the cheapest disks they can get away with so the better drives (NAS grade drives for example) that you would by for yourself are binned differently to the ones they use, so the failure rates aren’t really representative of all the disks made by a particular manufacturer.
I’d say the charts give a good general impression of stressed drives.
For example, seagate drives perform less well, but cost less.
As long as One has redundancy, one can use this.
Backblaze themselves basically say they are happy to buy a load of cheap drives, because they can afford for them to fail.
All drives fail, and all manufacturers claim millions of hours between failures.
I like WD drives for middling reliability, and a reasonable RMA process.
I have a few HGST drives, but haven’t tested their RMA, so can’t comment
Nope. It shows the weaknesses of a particular model, even a particular run of that model and that’s it. People always want to extend the data and it doesn’t work. BB’s stats are good enough because they have a tendency to buy in bulk which will give you plenty of data points for their system(s).
1 Like