Cashback brings the price to just under $560 - absolutely nuts for this much SSD storage.
Was looking at potentially replacing some of the HDDs in my server with these but still too expensive for my tastes, will have to wait another few years :P
Original PCOMP 20% off Sitewide at Computer Alliance on eBay Deal Post
It's just a home server for Emby, backups, etc.
Is QVO reliability not so hot?
Yeah, they're not as reliable by design.
These things best use case would be to run games off of.
I wouldn't store anything important on them long term.
Slightly misleading on the "not as reliable by design". They are not designed for a lot of writes to the drive. The more data they cram into a cell (SLC single bit, MLC, 2 bits…this one, QLC, is 4 bits) makes the cell less reliable to be written to - it wears out faster. E.g. if you just fill the drive once, then never write to it again, it will be basically as "reliable" as say a WD Green or 860 Evo SSD - basically. Reading data for all intents and purposes doesn't wear out the drive.
These ARENT designed to be in enterprise servers where they're getting hammered by terabytes of writes per day. In a home NAS where you're storing your holiday photos, these will be fine (ideal/perfect? no but fine).
NOTE: In ANY SERVER you should always have a backup of the data - personally anything non-critical (replaceable data) I use RAID5 as a "good enough" backup (note this IS poor form, but it's just a lot cheaper). For my critical stuff, photos, code, etc. I use a full mirror backup (e.g. Windows Backup). No media we have today - outside of tape (or possibly that M-Disc tech…) is reliable for long term storage. SSDs, USBs, HDDs, CD/DVDs, they're all prone to sudden failure and should be backed up.
@incipient: Samsung did do a decent amount of over-provisioning for 860 QVO, so it is fine to write regularly to these SSDs. However, you should save at least 10% free space to give SSD enough blocks to rotate the data writes.
I got burnt by Samsung 840 and 840 EVO with both suffering from slow old data reads. The dodgy fix was to refresh the data regularly. So, I wouldn't count on QLC data store won't degrade over time just yet.
The biggest issue with QLC SSDs is performance. They need very aggressive SLC cache to hide their true slow speed. Intel's D5-P4320 SSD is an enterprise SSD using QLC.
Backup movie files on an SSD? You must be rich.
$560 to backup 4TB of movies.
Notice the comma between Emby and backups…
@deanylev: Same - you must be rich. Jokes aside, SSD is not a good option for backups. Cost bites you two ways, it is already more expensive to start with and to minimise cost, you would go QLC, which lacks the sustained write speed. Also, a proper backup means you do more writes than reads (you backup often, but you don't restore every week) and that's not what SSD is suited for.
Media server for playback, SSD's usefulness is also questionable. Even 4K video playback doesn't need that much speed (you watch a 4K movie over 90-180 minutes, not 5 minutes). Unless you have a top class gaming setup, or you really need fast, burst reads very often, 4TB SSD is an overkill. However, if it is just loose change for you, then obviously go ahead.
@netsurfer: Definitely wouldn't be using the SSDs for the media server part, would just be for backups. Mainly for reliability, but yeah probably silly when I could just put HDDs in RAID. I don't even have a 10Gb network so wouldn't be able to take advantage of the faster speeds!
In this test Samsung 840 Series 250 GB lasted 900 TB of writes, which is about 10 years of filling the entire drive every single day. Older drive and backups are always a good idea, but if that's not "long" and "high traffic" I wonder what is for consumer grade use.
Pity my Samsung 840 250 GB lasted only 11 months, with only 1.2 TB total writes. It also suffered from the infamous slow old data read issue. I don't believe in these aggressive write tests in a short timeframe. Have 4 SSDs died so far, more than HDDs (1) died (and I managed to recover the data on the HDD with a board swap, unlike SSD - zero data recovered). None of the SSDs that died even reached 10% of the max endurance rating.
what are you doing with these SSDs that are dying so fast?
Most people SSD have lasted at least 5 years or more.
@pinkybrain: What do YOU think I was doing with SSDs which didn't even manage 1.2 TB total writes?
The answer is not much. My points are:
I have SSDs for MORE than 5 years and I have quite a number of them. You expect them to never fail? If you read Anandtech earlier on for SSDs when they initially started becoming available for consumers, you would know Anand (while he was still at Anandtech) stated: proceed with caution. Earlier on, SSDs had shockingly bad failure rate.
For those SSDs of mine which died, they all have been idle / unused for weeks and then I turned the PC or laptop on to use it, then it just died. Interestingly, heavily used SSDs (and used regularly - even with sector reallocations) seem to last longer for now (touch wood).
@netsurfer: I have no IDEA what YOU would be doing with the SSD that cause it to die early
hence why I asked. There's no way for me to know what your USE case is for these SSDs.
No need to ask a rhetorical question and giving an attitude
Also I miss the part where you wrote 1.2TB write.
So what are all the model that were the ones that died for you?
so they were all boot drives in different pcs and laptops?
Interesting that yours died quickly..
My Crucial M4 is still running as my boot drive.
Maybe cos mine use MLC nand, that is why it has lasted for so long.
@pinkybrain: Samsung 840, an SSD that came with a mini PC (it uses Samsung MLC NAND, but not the toggle NAND type - cheaper one), a MLC SSD from OCZ (yeah, I know - anyway, got a full refund for that), another TLC SSD using Samsung NAND. Other than 840 - which is used as a portable drive (which did have OS installed), all the others are system drives for non-critical systems.
While I prefer MLC SSDs (for better sustained performance and generally longer track record), I did have MLC SSDs failed. I still have a Crucial m4 (obviously, it needed a firmware update), but it hasn't been used much. I have even older Crucial SSDs (which according to SMART) the wear level is dropping and the cell quality has degraded to 60%.
My main point is this: don't go and tell people that SSDs are far more reliable and will last more than 5 years. Tell them that SSDs are just another storage type, do backup regularly. In fact, if you opted for first gen of a new SSD technology, there is no proven track record so there is risk. This one is the first gen QLC from Samsung. After being burnt by their first gen TLC, you cannot blame me for being cynical about reliability. I hope Samsung learned their lessons and did test these properly.
P.S. m4 has a decent record, but m4 is not the first MLC. There were a lot of MLC SSDs before with higher failure rate. Don't try to generalise it. Backup often and don't assume. Don't try to convince me to tell people SSDs are more reliable. It is not going to happen, not for a while. 4 SSDs failed is a fact (+ 3 of them now in degraded state; with 1 of them being an MLC which required a sector reallocation within 3 months - it had more than 50% free space ALL the time and very few writes).
@netsurfer: how do back your SSD?
do you regularly image the entire SSD to an image file?
Most of us use SSD for boot drive
so regularly imaging the SSD seems like the only option if we want to preserve everything.
@pinkybrain: You wrote:
what are you doing with these SSDs that are dying so fast?
I have no IDEA what YOU would be doing with the SSD that cause it to die early
Doesn't that imply you believe the issue is me not SSD? You started this. Since you started this, man up and admit it.
Also, read: https://photographylife.com/m-2-nvme-drive-overheating-failu…
Most people SSD have lasted at least 5 years or more.
Also I miss the part where you wrote 1.2TB write.
Your words, not mine. If you missed my comments, it is MY FAULT? Go and read some SSD failure reports, SSDs do fail the first year. If you go to Amazon, NewEggs, you would see people rating some quality SSDs 1 star due to really early failure.
Think about these:
The OCZ SSD which failed was known to have a failure rate above 50% so it was no surprised it failed (and obviously, I was expecting it - the retailer highly recommended a full refund, instead of replacement). So this MLC are more reliable is not necessary true (some were pretty awful earlier on). Samsung 840 was a disappointment. The other two turned out to be below expectation.
I take wear level into consideration. Does your 5 years cater for wear level drop? I can pretend my 3 degraded SSDs are still 100% - based on manufacturer's testing software. However, that's just not true, they are degrading (wear level dropping and/or already had bad sectors).
Most of us use SSD for boot drive
It's actually not that part that's the issue, I can easily re-install any OS quickly onto a new SSD (or perform a restore). It's that we also use SSD for in flight project work files which require fast read/write. Backing them up regularly is important (or for work, just check into the repository / source control server regularly). I have more than 20 SSDs in service so you don't need to tell me what to do.
@netsurfer: LOL, start what, admit what?
I just wanted to know why your SSD was dying so early ie. lasted only 11 months
since most SSDs are at least known to be more reliable than 11 months.
No need to "Make a mountain out of a molehill"…
If it is not your fault then so be it..
In regards to the 1.2TB comment, did I say it was YOUR FAULT?
I was just giving the context that I missed that comment.
Just chill out dude… or else you're going get an aneurysm.
I just wanted to know why your SSD was dying so early
You need to go back to the beginning chain, the review was using 840 SSD, and that was before the the slow read issue with 840 and 840 EVO was found. It was because that report was based on 840, that's why I commented.
Part of the reason for pointing out / showing Samsung 840 being a bad SSD was to highlight the fact that SSD makers do make mistakes. If you gone back to the history of SSDs, pretty much all makers made mistakes (inc. Crucial). The way Samsung handled 840 in particular was not professional at all. At lest two firmware fixes were attempted for 840 EVO, but zero for 840. While other SSD makers couldn't match Samsung on planar TLC in terms of speed for a long time, at least they didn't rush their products.
I've been trying to point that out a few times, but you kept on posting comments like:
I have no IDEA what YOU would be doing with the SSD that cause it to die early
The point was 840 had multiple issues so despite the bling bling review on its over-provisioning, it doesn't change the fact that it is a bad SSD - one of the mistakes from Samsung.
Basically, you could go and get an SSD that doesn't yet have a proven track record yet, but there is a risk. Even Samsung brand doesn't mean they are fine (nor any other brand).
Nice! This is what I've been waiting for thank you :)
I don't how how heavy your server load is, but these wouldn't last long in a high traffic server.
Definitely not their intended use.