Report says Microsoft will require SSDs for new OEM PCs soon

[Beerzerker]

16 hours ago, StDragon said:

Cost per GB, HDD are the way to go primarily for SAN / NAS units. But no one should be using an HDD as a the boot volume nowadays. And that last part is why MS is pushing SSDs, because the user experience is far superior compared with HDDs.

As for reliability, that depends how cheap the OEMs source their drives from. I've seen crappy Toshiba drives in Dell OptiPlex units take a shit several times (as in, having to RMA the drives) whereas something from Samsung is rock solid reliable. Yes, you're going to pay a premium, but options do exist. So, it's not an issue with SSD tech in general as it is with manufactures cheapening out on quality components.

 

10 hours ago, Zodiark1593 said:

The performance a pretty big issue with SSDs. A lot of ordinary people raise an eyebrow when finding their smartphone and tablet outperforms their PC for their day to day tasks, and the root issue is generally a slow storage drive. So sales is definitely a given. 
 

At any rate, counting on the possibility of recovering data from a failed drive is a very perilous mindset. Proper protection of data involves multiple duplicates in separate medium. 

I guess it's a matter of what's important to you, the user.
I've had failures from SSD's that were less than 6 months old (No less than three) and that's not all but so far, only 2 HDD's have ever failed from actual age.
I have no problem using an HDD as a primary drive or for any other use but my own needs are mine alone and I'm happy with them. 

[Zodiark1593]

57 minutes ago, Beerzerker said:

 

I guess it's a matter of what's important to you, the user.
I've had failures from SSD's that were less than 6 months old (No less than three) and that's not all but so far, only 2 HDD's have ever failed from actual age.
I have no problem using an HDD as a primary drive or for any other use but my own needs are mine alone and I'm happy with them. 

To be fair, if you run a tight ship, a HDD is perfectly serviceable for a boot drive. Not especially speedy, but not dog slow either. I've done it. And auto defrag worked a treat on Win 7.

 

However, even something simple like keeping startup programs down to a minimum is generally beyond the capability of most laypeople. Most OEMs get some sort of revenue stream from pre-installed applications, so I don't expect this to go away. SSDs go a long way to alleviating the performance impact through brute Random IO performance, allowing for a decently quick system out of the box. Even inexpensive drives often have enough overhead where even some pretty boneheaded naive users (the type likely to be running lots of startup programs, and even adware) are unlikely to run into significant slowdowns in their day to day use.

 

Additionally, Windows Update is drastically more tolerable with SSD rather than HDD. On the work systems, performing updates rendered the PCs useless for ten minutes or so (IE, the company actually loses substantial money for workers that are waiting on updates), while on my own SSD equipped systems, it's barely an inconvenience. Barely slower than doing an actual restart. Virus scans, similarly, are also very fast, being limited by the CPU nowadays.

 

Given the quality of life upgrades an SSD provides for even performing system maintenance, I wouldn't go back even if some gunman held a gun to my head with one hand, and a hard drive with the other.

 

Out of curiosity, what's your sample size for drive longevity? 3 drives from three separate brand seem rather extreme, unless these were some bootleg drives or something of the sort. I've 6 SSDs in use here. 3 NVME drives (660P, WD SN750, ADATA Swordfish), and 3 SATA drives (PNY Optima, AMD Radeon (back when they rebranded drives) and Patriot Flare). The oldest is the PNY drive, picked up in 2014 to replace the failed hard drive in my laptop (from 2010). Incidentally, that is the only Hard Drive, or really, any drive, failure I've had. The laptop is still in use today by the person I gave it to. Granted, my sample size is quite small.

[leadeater]

50 minutes ago, Zodiark1593 said:

Out of curiosity, what's your sample size for drive longevity? 3 drives from three separate brand seem rather extreme, unless these were some bootleg drives or something of the sort. I've 6 SSDs in use here. 3 NVME drives (660P, WD SN750, ADATA Swordfish), and 3 SATA drives (PNY Optima, AMD Radeon (back when they rebranded drives) and Patriot Flare). The oldest is the PNY drive, picked up in 2014 to replace the failed hard drive in my laptop (from 2010). Incidentally, that is the only Hard Drive, or really, any drive, failure I've had. The laptop is still in use today by the person I gave it to. Granted, my sample size is quite small.

I've had a few SSD fail. 2x SanDisk SSD Plus 120GB used as write-back cache in one of my servers and one very suspect Samsung 840 EVO that randomly started working again (I think, not had time to really test it).

 

One of the only real advantages to HDDs is even the lowest end cheap ones have the same typical reliability as more expensive ones, the distribution range between the best and worst is small. SSDs on the other hand when you go cheap like I did, on purpose for that server, you do actually get inferior quality product and I find it's the controller that dies. We have not reached that design point where we have for HDDs of that quality and reliability floor. We have to remember HDDs weren't always as reliable as they are today, not by a long shot. The first HDD I used used was like 4MB.

 

None of my 3 Samsung 840 Pro (1x 256GB, 2x 512GB), 4 Samsung 850 Pro 512GB have failed. I'm sure I also have another Samsung SSD in my parents laptop I swapped out the HDD from but I forget if it really is a Samsung or something else.

 

If include the SSDs from work then I can think of only 3 failures of SSDs in 8 years and none from NAND wear out.

[wanderingfool2]

This is a good thing...but I do hope that Windows will have prompts in it to prevent someone from thrashing their SSD.

 

While SSD's can be quite reliable it's super easy to kill an SSD especially a cheap one.  Specifically what I'm expecting to happen, people plug in their phone to transfer photos.  There is the convenient "pictures" in the Explorer quick access, so they naturally just store it there.  They fill up their SSD to 95% capacity, and have enough activity on the C drive to wear out the remaining drive.  (Or worse, getting it 99% full).

 

I've lost at least 2 SSD's that got demolished by Ubuntu because video files were being automatically saved to them (and the system let them get to the point of a few MB left).

 

Lots of scenarios like that...although now I think of it W11 is suppose to reserve a certain amount of space for Windows updates...so maybe that will be enough from preventing irreparable harm.

[leadeater]

3 minutes ago, wanderingfool2 said:

Lots of scenarios like that...although now I think of it W11 is suppose to reserve a certain amount of space for Windows updates...so maybe that will be enough from preventing irreparable harm.

huh unintentional overprovisioning, that's actually a nice point. I thought Windows 10 also started reserving space too after one of the updates or am I wrong?

[MageTank]

35 minutes ago, leadeater said:

huh unintentional overprovisioning, that's actually a nice point. I thought Windows 10 also started reserving space too after one of the updates or am I wrong?

Nope, you're correct. Has been a thing since 1903: https://techcommunity.microsoft.com/t5/storage-at-microsoft/windows-10-and-reserved-storage/ba-p/428327. 

[GoodBytes]

47 minutes ago, leadeater said:

I've had a few SSD fail. 2x SanDisk SSD Plus 120GB used as write-back cache in one of my servers and one very suspect Samsung 840 EVO that randomly started working again (I think, not had time to really test it).

 

One of the only real advantages to HDDs is even the lowest end cheap ones have the same typical reliability as more expensive ones, the distribution range between the best and worst is small. SSDs on the other hand when you go cheap like I did, on purpose for that server, you do actually get inferior quality product and I find it's the controller that dies. We have not reached that design point where we have for HDDs of that quality and reliability floor. We have to remember HDDs weren't always as reliable as they are today, not by a long shot. The first HDD I used used was like 4MB.

 

None of my 3 Samsung 840 Pro (1x 256GB, 2x 512GB), 4 Samsung 850 Pro 512GB have failed. I'm sure I also have another Samsung SSD in my parents laptop I swapped out the HDD from but I forget if it really is a Samsung or something else.

 

If include the SSDs from work then I can think of only 3 failures of SSDs in 8 years and none from NAND wear out.

So you are suggesting that this is made on purpose for people to lose their data and be forced/encouraged to use OneDrive backup which Windows shoves in people face?

 

You know... I was joking, but based on Insider builds, since a very long time, File History is partially gone. It's basically 1 step away from being removed.

 

Hmmm.. maybe some truth to this...

[Quackers101]

hate storage with mobile, never know how it works by the UI.

where you mix cloud storage with internal storage or memory card?

When trying to find it, it's hard, and sometimes you have to go through their cloud services anyways, which is kinda sad.

makes me like older software more and more, how direct your access was to the files you needed or folders and where they are located.

 

now it's all about searching through the web, cloud and services, not your "owned" personal space? I guess, or just have everything on an external drive.

[leadeater]

2 hours ago, GoodBytes said:

So you are suggesting that this is made on purpose for people to lose their data and be forced/encouraged to use OneDrive backup which Windows shoves in people face?

No? SSDs are still far better and failure rates of even the bad ones aren't enough to care about on a wider scale like this. As a consumer just buy/opt for better SSDs whenever you can.

 

At some point, probably fairly soon aka less than a decade, SSD failure should be at the same point as HDDs, literally not a factor at all, at least in these sense people discuss it now.

 

Problem for Microsoft is there is no good way to offer MS Accounts during setup etc without being accused of so many things etc. Having this option and actually using it is a good thing, having an off site copy of all your files and it actually being widely adopted is leagues better than now and the past where the vast majority of people had no data backups.

[cmndr]

2 hours ago, Zodiark1593 said:

To be fair, if you run a tight ship, a HDD is perfectly serviceable for a boot drive. Not especially speedy, but not dog slow either. I've done it. And auto defrag worked a treat on Win 7.

On Windows 11 with a 12100 and 24GB RAM:
think 5 minute boot time, 7 minutes to start a web browser and an inability to type into the browser bar and get responsive text.

It's entirely unusable these days. This is on a new Dell system. I didn't see anything egregious in the startup section.

 

 

 

1 hour ago, leadeater said:

I've had a few SSD fail. 2x SanDisk SSD Plus 120GB used as write-back cache in one of my servers and one very suspect Samsung 840 EVO that randomly started working again (I think, not had time to really test it).

CACHING is far more demanding than "normal use"

 

Part of the reason why I'm a bit of an optane fan (effectively unlimited writes) and am a dash sad that the technology will likely die.

[leadeater]

5 minutes ago, cmndr said:

CACHING is far more demanding than "normal use"

Oh I know, they were sacrificial SSDs on purpose lol

 

For that server I just wanted some SSDs, didn't care about good because I'd just swap them later for new larger cheap ones etc etc. The cache/hot tier because it's not just cache on that setup is mirrored SSDs so not a huge risk of data loss.

 

Think is has Adata SP550's in it right now, Adata something of some size. As you can tell don't care about that server so much lol.

[Kisai]

42 minutes ago, leadeater said:

No? SSDs are still far better and failure rates of even that bad ones aren't enough to care about on a wider scale like this. As a consumer just buy/opt for better SSDs whenever you can.

 

The thing I wish would happen is to pull a page directly out of Apple's playbook and backup the entire "OS+Application+Data" as an image that can be stored in the cloud. It's just not going to be a thing until more people are on 100Gbit fiber.

 

When I upload video streams to youtube, that's a good half day with a 15Mbit upload.

 

42 minutes ago, leadeater said:

At some point, probably fairly soon aka less than a decade, SSD failure should be at the same point as HDDs, literally not a factor at all, at least in these sense people discuss it now.

I'd make the argument that SSD's will never hit the "shelflife" durability of a HDD, because you can pull 3.5" drives with 40pin ATA connectors out of 386/486/Pentium systems and they will still work today. Where as Flash Media will almost certainly not because of how they physically work. So a HDD kept away from magnetic fields for 30 years, will still work. Flash Media that is not plugged into anything will almost certainly not. Go find PS1 memory cards and try to use them. They'll be 25 years old. Or MMC/SD/MemoryStick/etc cards from early 2000's. You'll be lucky if they still work because they were left in a device that was plugged in at some point. Bitrot and such.

 

 

 

42 minutes ago, leadeater said:

Problem for Microsoft is there is no good way to offer MS Accounts during setup etc without being accused of so many things etc. Having this option and actually using it is a good thing, having an off site copy of all your files and it actually being widely adopted is leagues better than now and the past where the vast majority of people had no data backups.

Honestly, there's three things they could do:

1) Separate OS(/Windows), Applications(/Program Files), and Data (/Users/***)

Backup the user's Data only. When installing a new OS, give the user the option to "install this OS with your existing data backed up to OneDrive"

or

2) Backup the OS+Applications+Data to "a new machine or USB device"

 

I know the latter has always been possible, as I've invoked it to work around a broken SCCM once, but it's such a disaster overall since any problems that OS has get transferred.

 

3) Store the Applications and Data for those applications as backup packages so the user could selectively restore these to a new OS without any of the data from anything else.

[LloydLynx]

On 6/9/2022 at 5:21 AM, Kisai said:

I can't tell you how much some of those dell latitudes suck even when they have a SSD in them.

[leadeater]

12 minutes ago, Kisai said:

The thing I wish would happen is to pull a page directly out of Apple's playbook and backup the entire "OS+Application+Data" as an image that can be stored in the cloud. It's just not going to be a thing until more people are on 100Gbit fiber.

 

When I upload video streams to youtube, that's a good half day with a 15Mbit upload.

Yea, OneDrive backup while nice isn't "everything". I think some settings are backup up but I think only the very new modern settings and only for "some things" which isn't good enough really.

 

I don't mind if it's not a backup you can recover directly from, as in fresh install recovery image, but applies once you connect up your account again and then it applies everything back. Being able to generate a System Image for download (wim file with media creation tool) would be icing on cake.

 

12 minutes ago, Kisai said:

I'd make the argument that SSD's will never hit the "shelflife" durability of a HDD, because you can pull 3.5" drives with 40pin ATA connectors out of 386/486/Pentium systems and they will still work today.

That's really not that important and fyi if all the cells zero out on the SSD due to not being powered for years it still works only the data is lost. SSDs aren't a offline long term backup medium and likely never will be. HDDs are a setup up, quite a big one, but can still fail from age even if not spinning.

 

Basically the only difference here is picking up a 15 year old HDD and being able to see what data was on is vs picking up a 15 year old SSD and maybe not being able to what data was on it, both will work.

 

12 minutes ago, Kisai said:

Backup the user's Data only. When installing a new OS, give the user the option to "install this OS with your existing data backed up to OneDrive"

This is how OneDrive Backup works right now, it just doesn't cover every setting but it covers all your files. It's just not enabled by default even when using a MS account to login. You just login and it detects if your files and setting are missing and then brings them down automatically.

[Zodiark1593]

46 minutes ago, cmndr said:

On Windows 11 with a 12100 and 24GB RAM:
think 5 minute boot time, 7 minutes to start a web browser and an inability to type into the browser bar and get responsive text.
 

That's pretty horrible, even by spinning rust standards. Unless something is very wrong, I can usually eke out a minute boot times, with another 30 seconds before the system is ready to go. This is on win 7, 8 and 10 machines. My Ubuntu system with the Pentium J2900 and HDD boots faster till. And my friend's laptop takes a couple minutes to get to desktop. I find it difficult to believe that Win 11 would be multiple times slower.

 

If you've weeded out startup applications, and you're not using one of those shingled drives, I'd be pretty suspect of that drive tbh. When I had my failure, things went slow, like glacial, for some weeks before SMART eventually kicked in (abnormally high read error rates), and I stopped using the machine until I picked up a new drive (my first SSD).

[cmndr]

2 minutes ago, Zodiark1593 said:

That's pretty horrible, even by spinning rust standards. Unless something is very wrong, I can usually eke out a minute boot times, with another 30 seconds before the system is ready to go. This is on win 7, 8 and 10 machines. My Ubuntu system with the Pentium J2900 and HDD boots faster till. And my friend's laptop takes a couple minutes to get to desktop. I find it difficult to believe that Win 11 would be multiple times slower.

 

If you've weeded out startup applications, and you're not using one of those shingled drives, I'd be pretty suspect of that drive tbh. When I had my failure, things went slow, like glacial, for some weeks before SMART eventually kicked in (abnormally high read error rates), and I stopped using the machine until I picked up a new drive (my first SSD).

My step father (who is a HUGE dell fan and who also refuses to get a $180ish refurbished i5 4500ish system because "refurbished") was pretty upset by how utterly unusable it was. His old (dead) 2010 i7 920 (upgraded to 6 core Xeon and a budget SATA SSD) system ran circles around it. His (currently a backup system) other computer could reboot (shutdown, post, load windows) and open the web version of Google Earth faster than the new system could go from windows desktop to a web page.

And yes. It was VERY unusable. Ubuntu on a flash drive is faster.

I suspect that some of it is just windows 11. I really didn't see that much crapware on it.
Cloning the HDD to an SSD completely solved all the issues. It's perfectly performant now.
 

[StDragon]

11 hours ago, Beerzerker said:

I've had failures from SSD's that were less than 6 months old (No less than three) and that's not all but so far, only 2 HDD's have ever failed from actual age.
I have no problem using an HDD as a primary drive or for any other use but my own needs are mine alone and I'm happy with them. 

Well, it would seem SSDs are more reliable than HDDs according to a Backblaze report. But there's not enough data to conclusively call it. In fact, the failure curve of SSDs over time almost matches that of HDDs; which IMHO is really fascinating given HDDs are electromechanical devices and SSDs are pure ICs.

[Beerzerker]

15 hours ago, Zodiark1593 said:

 

Out of curiosity, what's your sample size for drive longevity? 3 drives from three separate brand seem rather extreme, unless these were some bootleg drives or something of the sort. I've 6 SSDs in use here. 3 NVME drives (660P, WD SN750, ADATA Swordfish), and 3 SATA drives (PNY Optima, AMD Radeon (back when they rebranded drives) and Patriot Flare). The oldest is the PNY drive, picked up in 2014 to replace the failed hard drive in my laptop (from 2010). Incidentally, that is the only Hard Drive, or really, any drive, failure I've had. The laptop is still in use today by the person I gave it to. Granted, my sample size is quite small.

The 3 that failed were all purchased at the same time and were of the same brand and make, namely Munchkin  (Mushkin) Chronos drives.
I had bought 4 of them total and only one made it beyond six months, with the three that died going for about 2-3 months tops. I also had an older OCZ that died but it was a used drive when I got it, given to me and it did last a couple of years at least.
The lone Mushkin drive still works today and at least the 4 SSD's I bought later to replace those with (Corsair Force LS 60GB) are all still going. However even with that it's still at least 4 vs 2 HDD drives and one of those that finally quit (WD 30GB) was well over 10 years old and was used daily when it did crap out.
I was able to rescue what I needed before it was gone forever but from the SSD's that died, salvaging any data/files didn't happen at all with any of them.

[mr moose]

5 hours ago, StDragon said:

 according to a Backblaze report.

Yep, they have data sufficiently qualified to prove penis size is directly related to hdd failure rates, but not much else I am afraid.

 

Black blaze is going to require being bought out, all their staff replaced and peer review analysts from Nature over seeing all data acquisition and articles before I even consider anything they say to have any merit.

 

 

[Vishera]

24 minutes ago, mr moose said:

Yep, they have data sufficiently qualified to prove penis size is directly related to hdd failure rates, but not much else I am afraid.

 

Black blaze is going to require being bought out, all their staff replaced and peer review analysts from Nature over seeing all data acquisition and articles before I even consider anything they say to have any merit.

 

 

 

6 hours ago, StDragon said:

Well, it would seem SSDs are more reliable than HDDs according to a Backblaze report. But there's not enough data to conclusively call it. In fact, the failure curve of SSDs over time almost matches that of HDDs; which IMHO is really fascinating given HDDs are electromechanical devices and SSDs are pure ICs.

When it comes to HDD failures i am probably the luckiest man on earth.

The hard drives i use daily:

1 Year old TOSHIBA 4TB 5400 RPM drive (Don't remember the model)

3 Years old TOSHIBA P300 3TB 7200RPM

7 Years old Western Digital Blue EZEX 1TB 7200RPM

7 Years old TOSHIBA 2TB 5400 RPM (Don't remember the model)

9 Years old Transcend 1TB 5400RPM (Don't remember the model)

14 years old SEAGATE 80GB 5400RPM SATA (Don't remember the model)

17 years old TOSHIBA 60GB 5400RPM SATA (Don't remember the model)

 

Hard drives that died:

2011 SAMSUNG Spinpoint 500GB 7200RPM (Don't remember the model) - Died in 2015,Fried by a bad PSU that also killed the motherboard and the CPU

2013 TOSHIBA 500GB 5400 RPM (Don't remember the model) - Died in 2019 due to a mechanical failure.

[Kisai]

1 hour ago, Vishera said:

 

When it comes to HDD failures i am probably the luckiest man on earth.

 

I've had external drive chassis fail. All of them. Pretty much every external drive I bought that came in a chassis, failed other than the current WD models. Explicitly bought all the same/similar models so I would only have one power supply type to use. I suspect newer models may eventually just be USB-C, but at any rate. Every USB-2.0 external, going back to the Firewire model, I had to inevitably open the chassis and salvage the hard drive from, and all of them used cheaper WD Green or other 5400 RPM type drives.

 

Now it turns out that I haven't, personally, had any drive mechanically fail since the 90's. It's only been their external "shells", which is why I look at things like cheap NAS systems and kinda go "Nope, experience shows if I can't replace the power supply, it will die."

 

Speaking of which, that is what died in nearly every external, the power supply would just up and fail, but I only ever bought two of the same model once before, and sure enough when "that" drive failed, it's twin died within weeks.

 

So I have 4 external WD's that still work. The Apple Time Capsule? failure condition, it will spin up but then die after a few hours. Everything else I shucked the chassis and put in side the desktop until I replaced them with 7200rpm's.

 

The most unreliable drives are WD Green (which is what Apple used in the Time Capsule.)  Prior to the Quantum/Maxtor/Seagate mergers, Quantum was usually the "good brand", Maxtor was "the crap brand", and Seagate was not that better than Maxtor, with WD being the expensive-and-loud model to avoid. Prior to ATA drives, the previous MFM/RLL 34-pin drives were notoriously unreliable no matter who made them. Funny enough, the size reduction from 5.25" to 3.25" made them more reliable.

 

 

 

[Dr_Whom]

On the topic of reliability, speaking from a laptop repair perspective (mostly HP and Dell), I've seen a drastic reduction in drive replacements since the switch from mostly HDD boot drive to mostly SSD boot drives. And the SSDs that do fail seem to be controller related rather than actual flash failure. Strictly from the laptop perspective I think the fact that HDD's are susceptible to failure from shock damage (in a device that is meant to be portable) is a much bigger factor in "reliability" than the actual usage.

 

As far as what type of SSDs OEMs seem to be using, the two most common are your typical value SSD (cheap, DRAMless, 256-1TB, gets the job done), and these stupid Optane combo SSDs (I think these might be so common because of Intel incentives probably? No proof/evidence of that, but probably a good guess). It's an Optane module (usually 32 GB) and regular SSD (usually 512GB-1TB) on a single M.2 stick meant to be run with the Optane portion as cache for the SSD portion. These things by far have the highest failure rate from my personal experience and also have lots of issues with the cache link breaking just by looking at it funny.

[wanderingfool2]

11 hours ago, StDragon said:

Well, it would seem SSDs are more reliable than HDDs according to a Backblaze report. But there's not enough data to conclusively call it. In fact, the failure curve of SSDs over time almost matches that of HDDs; which IMHO is really fascinating given HDDs are electromechanical devices and SSDs are pure ICs.

Take what they post with a grain of salt though.  They don't exactly run things in the best of fashion, especially when you are comparing their HDD section.

 

I mean, iirc they got some of their harddrives by shucking...i.e. grabbing drives that very likely could have been dropped and damaged and using them.  There is also the issue that they left it on for 24/7.  While people do that to their PC's, I think it still is a bit of an unfair comparison between SSD's and HDD's.  Information can be gleamed from their published stats, but that involves sifting through their 30GB of log data (CSV files, one per day...got to tell you, it's a real pain to do that)