News New Backblaze Data Confirms SSDs Are More Reliable Than Hard Drives

[Admin]

Backup company Backblaze has upd its SSD lifespan data for Q2 of 2022, with some really intriguing results. The new Q2 data confirms that SSD life expectancy is much greater than hard drives.

New Backblaze Data Confirms SSDs Are More Reliable Than Hard Drives : Read more

 

[coromonadalix]

yeah right, lost tons of stuff on a brand new pcie 4 gen4 ssd after 4 months .... i'm sure and "classic hdd" would have survived / worked longer that that

do your backups regularly .... it saved my ass

 

[Jhfry]

yeah right, lost tons of stuff on a brand new pcie 4 gen4 ssd after 4 months .... i'm sure and "classic hdd" would have survived / worked longer that that

do your backups regularly .... it saved my ass

More reliable != Never fails.

That said, I wonder if their reliability figures are per device or per volume of data. For example, if SSD has a 1/100 failure rate vs 3/100 for HDD, but you need 4 SSDs to store your data, rather than 1 HDD, you have a higher chance of failure with your SSDs.

 

[cfbcfb]

When I can put one in the freezer for a while, whack it on a table and get it to run long enough for me to schlep my data off of it, that'll be the case.

That having been said, I still have every SSD I ever bought starting with a 64GB one, excepting that run of OCZ SSD's that desperately tried to implode the moment you used them more than a short while.

Those were hilarious. I put two of them in a raid mirror, was impressed by the speed until I saw it persistently faulting one of the drives and having to attempt to rebuild them on the fly.

 

[cfbcfb]

yeah right, lost tons of stuff on a brand new pcie 4 gen4 ssd after 4 months .... i'm sure and "classic hdd" would have survived / worked longer that that

do your backups regularly .... it saved my ass

The dirty little secret about gen 4 pcie drives, is that they all need a pretty good heat sink on them, or they run way too hot and thermal throttle, then the controller dies.

I use the sabrent one with the heat pipes on it, but I've also used those ones from amazon that you put rubber bands on. Those are really cheap and they work surprisingly well.

Even some of the gen 3 drives ran too hot. The Hynix Gold p31 was a rare exception. High nand density and extremely low power.

 

[watzupken]

yeah right, lost tons of stuff on a brand new pcie 4 gen4 ssd after 4 months .... i'm sure and "classic hdd" would have survived / worked longer that that

do your backups regularly .... it saved my ass

I don't know if you can refute the findings on the basis of 1 failed SSD. Of all the SSDs I purchased over the last decade, none of them have failed me. The oldest is the Intel G2 160GB which is still alive and kicking at 90+ % health for the wear indicator even after using it a lot when I got it over a decade back. On the other hand, out of the handful of mechanical drives I've purchased/ used over the years, I think at least half started acting up, or developing bad sectors. The problem with mechanical drives in recent years is the chase to the bottom in order to reduce cost. Good mechanical drives are hard to come by now because the quality is very poor. Enterprise grade drives should be better in quality, but I believe the quality have also dropped significantly over the last decade when SSDs started to replace them.

 

[USAFRet]

yeah right, lost tons of stuff on a brand new pcie 4 gen4 ssd after 4 months .... i'm sure and "classic hdd" would have survived / worked longer that that

do your backups regularly .... it saved my ass

Of my last two dead HDDs....1 died at 7 months, 1 died at 5 weeks.
So....

Have good backups, no matter what type of storage device.

 

[twotwotwo]

I'd love to see a repeat of experiments like this one to see how much write traffic current-gen drives will typically survive. I'm sure the TBW ratings are related to reality, but they're also aiming for a spec that 99%+ of drives will meet, since every drive that falls short is a potential warranty return. The average number of TB written before failure will be higher, and it'd be interesting to know how much higher.

 

[KyaraM]

yeah right, lost tons of stuff on a brand new pcie 4 gen4 ssd after 4 months .... i'm sure and "classic hdd" would have survived / worked longer that that

do your backups regularly .... it saved my ass

And not a single one of my SSDs ever failed on me so far, and I had a couple over the years with 6 of them active currently and some others that I didn't start up in a while since they are installed in old systems I didn't boot in a while. They were still healthy when the system was deprecated, though. 2 of them are over six years old and still quite healthy, which is surprising considering that one of them runs really hot.

Now, I have a 100% survival rate for my SSDs. However, that doesn't mean I don't expect any of them to ever fail, or think that this makes SSDs the most reliable storage devices on this planet. Maybe you should start looking at the bigger picture, too, instead of jumping the gun over a single failure you had. That sucks, no question, but since you were smart and backed your data up, what did you actually lose?

 

[AgentBirdnest]

yeah right, lost tons of stuff on a brand new pcie 4 gen4 ssd after 4 months .... i'm sure and "classic hdd" would have survived / worked longer that that

do your backups regularly .... it saved my ass

On the flip side... I own six SSDs that are 2-5 years old. The HDD I bought 2 months ago just died.

Luckily I do regular backups as well. 2 external HDDs, optical discs, cloud storage, and a data-swap with a trusted friend on the other side of the planet (they downloaded my files and keep it on an external drive, and I do the same for them.)
No, of course I'm not obsessive. 😛

 

[JamesJones44]

I wonder what the numbers look like when you separate by type (SLC, MLC, TLC, QLC, etc.). If SLC is weighting down the average then I can't say I'm very surprised, would be curious to know the spread of flash type used in the study.

 

[jonathan1683]

I have never had a SSD fail only HDDs, only 1 WD HD has failed on me and it was being used for metadata so it was being taxed a lot, moved it over to an SSD. I just stick to quality manufacturers like Samsung and WD, every Seagate/Maxtor I purchased failed.

 

[Johnpombrio]

3 years ago, I decided that I was done with spinning hard drives and bought 3 4TB Samsung SSDs, 2 2TB Samsung SSDs, and upgraded three NVMEs to 2 at 1 TB (Intel of all things) and one Samsung PCI 4 2TB more recently. Cost me a lot but Oh so much quieter. I use my old 4 TB hard drives as backup drives using a high-speed USB 3 dock to incrementally backup my 4TB SSDs and am contemplating a 2 TB SSD to clone my 2TB PCI 4.0 ( a hard drive takes way too long). That was my big push and now I have this huge case with no HDD racks in it leaving half of the case empty.
Over the years I have purchased at least a dozen SSDs of various sizes starting with 256GB none of which has failed. Contrast that with at least 5 hard drive failures in the two decades of using them of various brands. I certainly don't miss the "click of death".

 

[waltc3]

It's important to note that in this case we are only speaking of boot drives, which tend to be smaller than dedicated storage drives, and that is especially true of SSD boot drives. But you could use a 4TB SSD as a C:\ boot drive provided you partitioned it off, which is what I do with my PCIe4 SSD boot drive, a 500GB Samsung 980 Pro--I partition C:\ as ~225 GB and the rest is partitioned under another drive letter. I do it this way because all I keep on C:\ is the OS, utilities, device drivers, and related stuff. Makes it much easier and simpler to use Macrium Reflect to regularly back up C:\ after OS build upgrades, new program and game installs, etc. When I have everything I want on C:\, I've still got ~60GBs free on C:\. Much faster to back up 225GBs as opposed to a C:\ weighing in at a few TBs, and so on. I keep three C:\ incremental OS build backups at all times on a dedicated HDD partition. Works well for my system at home.

Where spinning drives always win is in raw storage, of course. My next data drive will likely be an 8TB + 3.5" drive--as atm I have 5 HDD's and two SSDs totaling about ~8 TB in total--these are HDD drives I've had for a while that I bring with me every time I upgrade. Still running fine, no problems, yet.

My PCie4 Samsung 500GB 980 Pro runs great full bore PCIe4 and I have no overheating problems with it, but I can't say the same for my 250GB PCIe3 Samsung 960 EVO .M2--with the 980 Pro I can do a full Defender AV scan of C:\ running flat out with no troubles, from start to finish. But with the 960 EVO, I cannot finish a full C:\ AV scan before the heat throttles the drive to a full stop! Interesting, because although the 980 Pro sits in the first M2 slot, directly beneath my 6900XT, the 960 EVO is in the second slot with nothing installed on top of it. And, with both drives, I use only the very thin metal "heatsinks" (if you will) that came with the motherboard...! Maybe 500GBs with the 980 Pro, even at PCIe4 speeds, is the sweet spot for that drive--and that little heatsink does fine. But the 960 EVO is even smaller but overheats much faster running at the slower PCIe3 mode.

I can definitely see how PCIe4/5 TB-sized M2 SSDs could require a more pronounced heatsink than what my 980 Pro presently uses! Interesting thread, guys...😉

 

[cfbcfb]

I don't know if you can refute the findings on the basis of 1 failed SSD. Of all the SSDs I purchased over the last decade, none of them have failed me. The oldest is the Intel G2 160GB which is still alive and kicking at 90+ % health for the wear indicator even after using it a lot when I got it over a decade back. On the other hand, out of the handful of mechanical drives I've purchased/ used over the years, I think at least half started acting up, or developing bad sectors. The problem with mechanical drives in recent years is the chase to the bottom in order to reduce cost. Good mechanical drives are hard to come by now because the quality is very poor. Enterprise grade drives should be better in quality, but I believe the quality have also dropped significantly over the last decade when SSDs started to replace them.

There was a rather lengthy run of drives from OCZ that failed while still very young. I lost a few. There have also been some defective drives and/or ones that couldn't handle heat very well. Sata SSD's have fewer of these problems overall, but I find the heat management on gen 4 and presumably gen 5s to be a real problem. That is pushing on one side, while the other side is being squeezed by Q flash drives that may in fact die during a normal drive life span of 5 years, due to low TBH and flash failures.

So while I agree that a drive with no moving parts should be better, I'm also noting that some sub standard flash and controllers are out there, and a good backup is ALWAYS a great idea.