I have personally been responsible for about 20 HDDs in the last 20 years and had about four of them fail before being put out of service for some other reason (1 in 5) but out of about 7 SSDs I have had zero failures. The pool of zero failure drives is much larger if I lump together SD cards with SSDs.
This is a bad rule of thumb, because NAND Flash is a physical device with moving parts - cells grow and shrink during normal operation, this is where the limits on cell erase cycle count comes from, worn out cells develop cracks and leak electrons.
An SSD may be more reliable than an HDD, but when a drive does fail during normal operation[0] it appears that data recovery is usually significantly more feasible/cheaper with an HDD. I’d be glad to be proven wrong.
See also Google’s paper[1]. It doesn’t seem to be especially black-and-white.
[0] This excludes cases such as e.g. dropping an HDD during operation and ruining large plate areas, of course.
In theory an SSD failure should always be more recoverable: eventually they should be put into a read-only state, and then data can be pulled off easily.
In practice, it's a bit more nuanced since SSD failures aren't always the "we wrote too many cells, this block is now read only", and tend to be more potential other issues (driver malfunction, controller chip issue, power hardware, etc).
With an HDD failure, if the hard drive itself is damaged/end of life/whatever, the act of reading it can cause further damage, so in a graceful failure the SSD seems to win.
I'd suggest that the cost of data recovery on a HDD exceeds the practical limit for what most consumers would be able to stomach (since data recovery companies tend to charge thousands...) even for HDDs. Hence why redundancy/backups/RAID are important, I guess.
In practice I dont know of a single SSD brand/model that goes read only upon failure. I do know specific cases of manufacturer lying about this ability tho, like Intel.
But almost nobody recovers data from a broken HDD other than personal computers and even then only in exceptional circumstances. The best protection is still to have a backup. In my many years of work in IT I have never heard of any recovery from any drive other than a personal laptop (and that only once).
I recall a drive that got "sticky". It would appear dead, but if you shook it clockwise and counterclockwise around the rotational axis, it would get "unstuck" and work again.
I figured this out after the first time it happened, when I mailed it to be fixed and it was returned working but of course my data was wiped.
Even assuming this is true, typical recovery is order or two of magnitude of the cost of the drive. So it is still small percentage of drives that receive any treatment in case of failure.
I think it isn't helped by the fact that SSD fail in a more specific way, you can write x amount of bytes before it should fail, so it feels more like a countdown to failure when you use one even though you could probably never write that much in your lifetime under normal loads.
'Study' is a strong word for what's being reported.
> Keep in mind that the average age of the SSDs Backblaze uses is only 12.7 months old, whereas the average age of its hard drives is 4x older, 49.6 months to be precise. Also, Blackblaze uses SSDs as boot drives alone in its servers, so these SSDs could also have less of a workload compared to the actual hard drives where are constantly being used to backup client data.
I think this is a goof in Tom's Hardware's reporting. In the Backblaze report, it says:
"What we have are two groups of drives, one HDDs and the other SSDs, which have performed the same functions in the same environment over time. The table below compares the failure rates in aggregate for Q1 2021 of our HDD and SSD boot drives."
So while the age is different, the function is the same for the HDD/SSD comparison section of Backblaze's report.
Considering how close the NAND / SSD are to CPU in most Laptop and MacBook. I wonder if this could be a problem in the future. Especially recent MacBook which has no cooling Fan.
There are several conditions that must happen before data loss begins due to cell corruption when the drive is stored (unpowered), with abnormal temperatures being one of them. The primary condition is the drive must have reached its end of life.
23 comments
[ 2.3 ms ] story [ 61.6 ms ] threadThis is in line with what they observe.
See also Google’s paper[1]. It doesn’t seem to be especially black-and-white.
[0] This excludes cases such as e.g. dropping an HDD during operation and ruining large plate areas, of course.
[1] https://arxiv.org/pdf/2012.12373.pdf
In practice, it's a bit more nuanced since SSD failures aren't always the "we wrote too many cells, this block is now read only", and tend to be more potential other issues (driver malfunction, controller chip issue, power hardware, etc).
With an HDD failure, if the hard drive itself is damaged/end of life/whatever, the act of reading it can cause further damage, so in a graceful failure the SSD seems to win.
I'd suggest that the cost of data recovery on a HDD exceeds the practical limit for what most consumers would be able to stomach (since data recovery companies tend to charge thousands...) even for HDDs. Hence why redundancy/backups/RAID are important, I guess.
You are right.
But almost nobody recovers data from a broken HDD other than personal computers and even then only in exceptional circumstances. The best protection is still to have a backup. In my many years of work in IT I have never heard of any recovery from any drive other than a personal laptop (and that only once).
I figured this out after the first time it happened, when I mailed it to be fixed and it was returned working but of course my data was wiped.
https://youtu.be/99Sk2ZnjGQA
Even assuming this is true, typical recovery is order or two of magnitude of the cost of the drive. So it is still small percentage of drives that receive any treatment in case of failure.
Also market cap != amount of annual business
>almost nobody recovers data from a broken HDD other than personal computers
> Keep in mind that the average age of the SSDs Backblaze uses is only 12.7 months old, whereas the average age of its hard drives is 4x older, 49.6 months to be precise. Also, Blackblaze uses SSDs as boot drives alone in its servers, so these SSDs could also have less of a workload compared to the actual hard drives where are constantly being used to backup client data.
"What we have are two groups of drives, one HDDs and the other SSDs, which have performed the same functions in the same environment over time. The table below compares the failure rates in aggregate for Q1 2021 of our HDD and SSD boot drives."
So while the age is different, the function is the same for the HDD/SSD comparison section of Backblaze's report.