SSD drives, for boot and use .

[tangor]

Just spent a few days (personal machine) replacing a 240gb SSD that worked really well, but only lasted 1 year and 11 days. PITA, of course because you build your personal system different than work, you live in it, and having to start over is like getting a new pair of shoes and having to go through the time and pain of breaking them in. If this was a one off I'd go with the flow and not worry, but, having had the same problem with a few servers I have and for some I manage, I've now gone back to spinning rust (best possible of course) for the reliability.

What have your experiences been with mission critical SSD in the workplace? I do love the speed and the lower energy costs, but if the dang stuff will only last a year I'm losing money (and more urgently having to do image mirrors of such drives on a very constant (twice a day or more frequently) and that is counterproductive to keeping expense and operations to a minimum. SSD could be the greatest thing since sliced bread, but you couldn't prove it by me: In the last 3 years six out of six SSDs used for boot drives have failed, on average, at 1.3 years each.

[ken_b]

Well that's discouraging.

[lammert]

In my experience all systems with SSD disks need tuning to get a decent life time. The first flash based systems I used almost 20 years ago could only survive 10 to 25 writes per block. Flash manufacturers claim much higher numbers of write cycles now, but if your system uses significant writes on a nearly filled disk, you should expect crashes due to flash memory cell failures.

Never put swap files or partitions on SSD for this reason for example. Database files which are primarily read can be on SSD, but highly volatile index files may be better of on a RAM disc and recreated every startup of the database. Nothing beats RAM in speed.

If a bit fails on a magnetic disc, only that sector is unusable. If a bit fails on an SSD however the whole disc may be at risk due to the dynamic allocation of virtual disc sectors to flash blocks.

An extra problem is that consumer grade SSDs store up to 4 bits in one flash memory cell. This makes them cheaper and with higher storage capacity on the same chip size, but reliability degrades rapidly. Always try to buy SLC drives which stores only one bit per cell and not the cheaper MLC drives. MLC-4 cells may start to fail after only a few hundred writes while SLC memory cells should easily survive 100,000 writes.

[tangor]

Reality is that (under many tests published) consumer grade SSD should (and do) hit the Petabyte write and rewrite stage with ease.

In my applications I have NOT used them for general write/delete purposes, only as BOOT disks and ease of mirror copy to maintain settings that do not change all that much.

But (sad to note, but out there) all SSD are subject to catastrophic failure by unexpected power outages. Even with UPS installed, the power supply itself can queer the deal when it fails as opposed to the power grid itself. Turns out that's near certainty with SSD and rarely with HHD.

Makes me crazy. :)

But, all things said and done, what have I missed for a BOOT ssd? About 40 seconds real time re: performance of one against the other. By going back to spinning rust I have a longevity I can't count on from SSD. A trade off, to be sure, but for the peace of mind ....

I want to love 'em, I truly do. Just can't at the moment.

[IanCP]

I use one SSD as the boot drive, Win 10 OS. I try to keep all programs on another drive though. Next desktop will keep as much as possible on another drive. A shame Windows wasn't more customisable.

Having said that, I'm not that particularly fussed about it. Fast booting up? I don't know, I push the on button and go and turn the jug on for coffee. Would I buy another SSD in the future? I dunno.

I also have seven drives in total.

[engine]

I can only speak from the perspective of a SSD in tablets and a laptop. So far, not a single failure in use, but, perhaps that's an exception.
Could the problem lie in the way Windows uses the SSD?

[jimh009]

I can only speak from the perspective of a SSD in tablets and a laptop. So far, not a single failure in use, but, perhaps that's an exception.
Could the problem lie in the way Windows uses the SSD?

Ditto for me - never once had a failure of any kind (knock on wood). But I'm using Apple products (Macbook, Ipad, etc..), so perhaps Apple SSD's are different?

Having experienced a fair number of problems with "spinning rust" as the OP lovingly called them, I don't miss those drives at all. They are loud, slow and drain the battery.

[J_RaD]

This is why i'm not widely adopting SSDs in my machines... customers ask all the time..and that is what i tell them FAILURE RATE IS HIGH 1 to 1/2 years and its toast.
The quality units last just fine...but they are $$$$$$$$$$$$$$ and once they see the cost, eh no thanks!

Its even still cheaper to RAID 0 two disks.

[tangor]

Laptop and tablet SSDs aren't subject to extreme power changes because these are battery operated devices. If your battery is low enough to cause a problem to the drive your device is already complaining "Charge Me Now!".

[engine]

So, the consensus appears to be that they are fine for tablets and laptops, but not so good for servers? I don't quite know why that would be the case because, as I understood it, it's the writing that causes the SSD to "wear out" whereas reading is fine. What is failing? Is it cooking, and it's the connection that's failing?

Laptop and tablet SSDs aren't subject to extreme power changes because these are battery operated devices.

I can't work out why the power would not be regulated on other systems.

Yes, i've had HDs fail, and some without warning, but, after the very first failure, my backup routine has saved me from too much pain when there's been failures. Of late, the failures have been minimal. Perhaps i'm on borrowed time with someofo the HDs.

[Brett_Tabke]

I have 6 SSDs here on boot drives, and zero failures in many years of usage.

I have a 256gig ssd that has windows 7 on it, that I installed in late 2009. Still running strong. It was used daily from 9 to 14 when I switched to new machine. It still is running all the time.

We have 12 win XP based registration computers for Pubcon that have SSD's on them and all are rock solid.

I've never had a SSD fail. We are going to all boot drive ssds here and moving all 2nd drives to big (3-8tb) spinners.

However, we have two 12bay NAS's with 3 tb spinner Seagates in them. We've had 7 drives fail in three years.
[extremetech.com...]

[J_RaD]

I can't work out why the power would not be regulated on other systems.

cause it is, you are correct this has ZERO to do with power fluctuations, it is with WEAR OUT.

OEMs are hand picking chips and getting great deals..so what goes in the OEM stuff should be fine.

Everyone on the outside world, we don't get great deals on the good stuff, you buy the quality stuff or you don't go to SSD.

hard drives in general, just like brett's link. Seagate quality is toast, kinda crazy for me to watch them go from super awesome in the early 00s to total junk.

HSGT , Western Digital, anyone else, seagate.

SSD, Intel, samsung, sandisk

[engine]

I know HDs can fail, so I use RAID arrays for backups and important data. If the PC crashes, no worries. I've had more problems with other components failing so the HDs i've been using have, so far, survived. The HDs in the TV recorders have been the least reliable, and i've lost a number of recordings through those failures.
Quality SSD prices can't come down soon enough for my liking, although, it'll be a long time before 1TB and above is affordable compared to HDs for every-day use.

[J_RaD]

The one good thing about spinning disks, once they pass around 1+ year... with no failure or hiccup, they normally will keep going till the end of time. I have some old drives with SERIOUS hours on them..and im talking like 10 years.. my old 10k rpm SCSI 320 drives are even still kicking.

[tangor]

Just to clarify my reluctance on SSD at the moment:

Each of the six drives mentioned were hit by unexpected power loss. Two from aging power supplies, three from a hurricane during which a car took out a power pole and a lightning strike took out a power station (any of which could have caused the outage), and one SSD lost by a stupid electrician killing power to the building. The three gone by hurricane were protected by a UPS ... but that failed when the battery died and no one was there to shut the systems down. Two were protected by UPS, but the system power supplies did the dirty deed. The last one was my home computer and that really ticked me off. :)

I've had no operational problems with SSD otherwise. But some of my systems are mission critical and having experienced this over a relatively short period of time I'm not willing to continue at the moment.

But I do miss the read speed to load some aps, besides the OS, such as Word, Excel and Access. All data is/was written to spinning rust, not the SSD(s).

[Andy Langton]

No SSD failures and multiple hard drive replacements in the last few years over here. I have no hesitation in replacing a failing boot drive with an SDD.

[J_RaD]

each of the six drives mentioned were hit by unexpected power loss.

an unexpected power loss should NOT result in a dead drive....what is this a server room or real life? all desktops require a UPS? so if the power blips for one second they all don't have dead bricked computers? STUPID.

spinning disks recover from power outages like this....press button...boot.. resume work.

if you test them with repeated power outages they might request a scan and fix....other then that, all good.

[Lugh]

I've had 2 boot SSDs fail in my desktop I built 17 months ago. Also had 1 HD failure.

Otoh I added a 2nd SSD to an Alienware laptop 3 years ago, and never a glitch from either.

My research says SSDs are very good these days, so I'll stick with them for OS and programs. Data goes on HD because of $/MB.

[J_RaD]

My research is, I build over 30+ systems a month, that I then have to stand behind.
Patterns quickly arise when this much hardware hits your bench... on my do not ever buy list OCZ! do not touch any SSD from OCZ!

I still firmly support spinning RAID 0 for speed, reliability, and cost.

[lammert]

@J_RaD, Did you only see a relationship between brands and reliability or also between SLC vs. MLC drives and reliability? The brand you mentioned seems to be active in the MLC+TLC market only.

[J_RaD]

for the most part it seemed to follow the brand, and the controller used.

[tangor]

Oddity.

Last SSD that failed me sat unplugged for a bit more than three weeks. Fiddling around with parts headed for the dust bin I plugged it in on a shell system and the damn thing booted. Color me perplexed. Back on the parts shelf ... but I won't use it mission critical for the time being.

[J_RaD]

I would say that drive has a suspect controller and will continue to be ghostly unpredictable.

When SSDs 1st started hitting the market everyone started W/R testing them... a bad controller would fail a drive every time, even within DAYS.

my guess is your powered down crap drive reset its controller, started playing nice...but given enough time will start behaving badly...so yes parts bin only.