Is SSD can die after period of time?

[amitabhbansal]

Friends i was going through with youtube videos and found one video defining the life span of SSD, i am really worried about it, is SSD are not much reliable then HDD?
here is the link of the video

 

[Mr.Blacky]

Did you even watch the video until the end?

 

[amitabhbansal]

I watched till end, but i am just asking is SSD is less reliable then HDD?

 

[yitwail]

I watched till end, but i am just asking is SSD is less reliable then HDD?

It depends on the SSD and the HDD, the capacity of the SSD, how much of the SSD is used, and available RAM. As video pointed out, all things being equal, the more free space available in the SSD, the longer it will last. I expect a 1TB SSD, which is what's in my MacBook Air, will easily outlast other parts of the computer, like the battery, keyboard, and display. I also mentioned RAM because swap memory is data that's written to the SSD, so if you're constantly low on free memory, then the SSD will wear out faster. In conclusion, get the largest SSD and most RAM you can afford. Regardless, HDDs do have the advantage that in theory you can replace them, if they break down, but that only matters if you're keeping the computer for a very long time, like 10 years or more.

 

[cmaier]

I’ve had 6 or 7 8TB Barracuda HDD’s die at 10000 hour uptime in the last few years. I’d much rather rely on SSD than on HDD when given the choice.

 

[rui no onna]

I've had both HDD and SSD failures. For HDDs, I have a box of failed drives from all brands (WD, Seagate, Samsung, Hitachi, Toshiba). If you care about the data stored, 3-2-1 backup rule.

For mobile laptop use, SSD is much better. I used to kill the 2.5" HDDs on my laptop around the 13-15 month mark. HDDs really hate being moved around while running. Spent I believe $600-700 on a 1TB SSD as soon as consumer ones hit the market so I could finally upgrade my laptop. SSDs are both faster and more reliable tha HDDs for my particular usage. Only downside was they cost 10x as much for same capacity at the time.

Granted, I can easily replace the SSD on my Windows PCs. Dead soldered NAND flash on the M1 MacBook Air would be much more annoying.

 

[leman]

Modern SSDs are generally more reliable than HDDs and also have longer expected lifespans on average. SSDs have limited amount of data that can be written to them before they fail, but these amounts are high enough that a normal prosumer user has no reason to worry (it might be a problem if you use it to run a production write-heavy database though).

Any electrical component can fail at any time. Backup your data.

 

[Ifti]

You'll probably find you replace the equipment before you have to replace the SSD.

 

[rui no onna]

Modern SSDs are generally more reliable than HDDs and also have longer expected lifespans on average. SSDs have limited amount of data that can be written to them before they fail, but these amounts are high enough that a normal prosumer user has no reason to worry (it might be a problem if you use it to run a production write-heavy database though).

Hmm, dunno about that.

If it were Windows PCs, yeah sure, I'd say SSD write endurance is well beyond the PC's useful life. I find my M1 MBA writes 20x more data per day than my Windows 10 PCs though and that's with just Safari browsing. Reckon the issue is the SSD is not user replaceable if it fails.

Any electrical component can fail at any time. Backup your data.

Too true.

 

[satcomer]

I watched till end, but i am just asking is SSD is less reliable then HDD?

Over the years SSD sense the full size SSD of first gen ration were the cheap ones that died about 30% of the time! After that period of time the manufactures seem to fix that problem! Sense the Zeros SSDs have been good unless you walk into a scam, so don't go starchy stores and use big box SSDs instead if you want reliability!

 

[leman]

If it were Windows PCs, yeah sure, I'd say SSD write endurance is well beyond the PC's useful life. I find my M1 MBA writes 20x more data per day than my Windows 10 PCs though and that's with just Safari browsing. Reckon the issue is the SSD is not user replaceable if it fails.

There was a bug with reporting written data on early M1 models but that was fixed months ago. Do you still see high writes?

 

[Spindel]

I’ve had 6 or 7 8TB Barracuda HDD’s die at 10000 hour uptime in the last few years. I’d much rather rely on SSD than on HDD when given the choice.

That's because you used Seagate drives, ever since the late 90ies when started to build computers (before I moved to Mac) my (anecdotal) experience was that if a drive failed that early you could bet it was a Seagate.

People stop buying Seagate products, they are crap!

EDIT:// I just had to check my NAS, my two 4 TB drives have had an uptime of a bit over 45 000 hours currently and my two 8 TB drives have had a bit over 20 000 hours. Crosses fingers that they'll run for at least 45 000 and 70 000 hours more respectively. And no they are not Seagate drives.

 

[cmaier]

I have two nas boxes each with 12 drives. Seagates work fine other than this one model. I’ve got seagate 6tb and 12tb with years of uptime. WD and hitachi work well too.

 

[Fishrrman]

All drives will eventually fail (of course).
Some will fail sooner than others.
SSD's will fail, just as do platter-based hard drives.
Some SSDs will fail sooner than others, with no warning.

I've had SSD's that suddenly "went dark" on me, never to "be seen" again.
Like nothing was there at all...

 

[Homy]

He says literally in the beginning "you don't have to worry and I'm gonna tell why" so I didn't watch the rest. ?

 

[jdb8167]

There was a bug with reporting written data on early M1 models but that was fixed months ago. Do you still see high writes?

There was never a bug in the reporting of total written bytes. The open source smartmontools always worked correctly. There were real bugs in the virtual memory of early versions of Big Sur on the M1. Most of those issues have been resolved if the complaining on these forums is anything to go on.

 

[Adarna]

Friends i was going through with youtube videos and found one video defining the life span of SSD, i am really worried about it, is SSD are not much reliable then HDD?
here is the link of the video

You're more likely to die first before your SSD conking out

 

[leman]

There was never a bug in the reporting of total written bytes. The open source smartmontools always worked correctly. There were real bugs in the virtual memory of early versions of Big Sur on the M1. Most of those issues have been resolved if the complaining on these forums is anything to go on.

The tools just grabbed the information provided by the Apple APIs. If I remember correctly, Apple claimed that these writes never actually happened, just the reporting was broken. No idea whether it is true or not. But M1 Macs have been on the market for almost one and half years now and we never saw reports of widespread failures, so if these writes really did happen than either Apples SSDs have extremely good endurance or the reported writes were indeed bogus.

 

[pshufd]

I've used SSDs since 2009 or 2010. Not one of them has failed. I'm still using some old 300 Mbps SSDs from the stone age. I imagine the stuff made today lasts longer than the stuff from a decade ago.

 

[mr_roboto]

The tools just grabbed the information provided by the Apple APIs. If I remember correctly, Apple claimed that these writes never actually happened, just the reporting was broken. No idea whether it is true or not.

I don't recall that at all. There was never any direct statement by Apple, just rumor sites reporting that an anonymous Apple employee said so. Which doesn't mean a thing, because guess what, rumor sites don't care very much about vetting stories.

The other thing is, even if that anonymous tipster actually existed and was accurately quoted, we know they were wrong! I tested M1 SSD behavior myself and found that Apple's NVMe SMART implementation reports host bytes written exactly according to NVMe spec. IMO, the whole idea of a reporting bug was just nonsense made up and circulated by fanboys on forums until it jumped over into rumor "reporting" and thence into your brain, where it lodged itself as if it was a fact, even though it was not.

The main cause of excess writes was an overeager swapout algorithm in the XNU kernel:

https://twitter.com/x/status/1396374313591140357

Some of Hector's other tweet threads talked about it affecting Intel Macs too, because XNU's VM code is not fundamentally different between Apple Silicon and x86. (The very lowest level details of how to manipulate page tables are different, of course, but policy decisions like how much memory pressure is enough to start swapping things out, and selection of what gets swapped out? That stuff's generic crossplatform code, as I understand it.)

However, as this bug was introduced with Big Sur, and Big Sur launched alongside M1, people's attention was on M1 and suspicion fell on the hardware even though it was really the OS.

 

[leman]

I don't recall that at all. There was never any direct statement by Apple, just rumor sites reporting that an anonymous Apple employee said so. Which doesn't mean a thing, because guess what, rumor sites don't care very much about vetting stories.

The other thing is, even if that anonymous tipster actually existed and was accurately quoted, we know they were wrong! I tested M1 SSD behavior myself and found that Apple's NVMe SMART implementation reports host bytes written exactly according to NVMe spec. IMO, the whole idea of a reporting bug was just nonsense made up and circulated by fanboys on forums until it jumped over into rumor "reporting" and thence into your brain, where it lodged itself as if it was a fact, even though it was not.

The main cause of excess writes was an overeager swapout algorithm in the XNU kernel:

https://twitter.com/x/status/1396374313591140357

Some of Hector's other tweet threads talked about it affecting Intel Macs too, because XNU's VM code is not fundamentally different between Apple Silicon and x86. (The very lowest level details of how to manipulate page tables are different, of course, but policy decisions like how much memory pressure is enough to start swapping things out, and selection of what gets swapped out? That stuff's generic crossplatform code, as I understand it.)

However, as this bug was introduced with Big Sur, and Big Sur launched alongside M1, people's attention was on M1 and suspicion fell on the hardware even though it was really the OS.

Ah, thanks for that, I was not aware there was more discussion. Was the bug ever documented? I saw the initial tweet by Hector but no followups to it. Do we really know that the data was actually written?

 

[pshufd]

Ah, thanks for that, I was not aware there was more discussion. Was the bug ever documented? I saw the initial tweet by Hector but no followups to it. Do we really know that the data was actually written?

Has anyone in consumer use had an SSD fail for too many write?

 

[leman]

Has anyone in consumer use had an SSD fail for too many write?

I am only aware of a single case (reported last year), which — if I remember correctly - involved using an M1 Mac to host a write-heavy database. I don't recall any other reports.

 

[pshufd]

I am only aware of a single case (reported last year), which — if I remember correctly - involved using an M1 Mac to host a write-heavy database. I don't recall any other reports.

I think that it was a bank running a Postgres application. They purchased consumer hardware to run an enterprise application. They should have purchased an enterprise-class SSD. The thing is that it was a mini so they could just plug in an external drive and keep right on ticking.

 

[mr_roboto]

Ah, thanks for that, I was not aware there was more discussion. Was the bug ever documented? I saw the initial tweet by Hector but no followups to it. Do we really know that the data was actually written?

Documented by Apple? Not as far as I know. Data actually written? Yes, as I said I tested the NVMe SMART reporting personally and it accurately reported the quantity of data I wrote, and most (all?) people reporting excessive writes were relying on NVMe SMART.