Slashdot Mirror
OCZ Toshiba Breaks 40 Cent Per GB Barrier With New Trion 100 Series SSD
MojoKid writes: OCZ is launching a brand new series of solid state drives today, dubbed the Trion 100. Not only are they the first drives from the company to use TLC NAND, but they're also the first to use all in-house Toshiba technology with the drive's Flash memory and controller both designed and built by Toshiba. That controller is paired to A19nm Toshiba TLC NAND Flash memory and a Nanya DDR3 DRAM cache. Details are scarce on the Toshiba TC58 controller but it does support Toshiba's QSBC (Quadruple Swing-By Correction — a Toshiba proprietary error correction technology) and the drives have a bit of SLC cache to boost write performance in bursts and increase endurance. The OCZ Trion 100 series is targeted at budget conscious consumers and users still contemplating the upgrade from a standard hard drive. As such, they're not barn-burners in the benchmarking department, but performance is still good overall and a huge upgrade over any HDD. Pricing is going to be very competitive as well, at under .40 per GiB for capacities of 240GB, 480GB and 960GB and .50 per GiB for the smallest 120GB drive.
And yet many other people have no problems running SSDs some people still using their originals.
Whenever I see someone say "had to replace them all" I can only think of device incompatibilities engineering screwups, or part selection screw-ups.
In short, there is no reliability issue, and the write limitation is a non issue for 99.999% of the computers out there. It just doesn't seem to be working for *you*
First and second gen SSDs were garbage, people are reporting 2 petabyte write lifecycles on them. Samsung just announced 10 year warranties on their consumer models. Intel has been offering 10 year warranties on their enterprise models for a few years now.
That said, if you bought anything other than Samsung prior to about 2013, the "old" OCZ in particular (the "new" OCZ is using the corpse of their brand name for Toshiba manufactured drives now) had failure rates in the 15-20% real world return rate numbers reported by retailers. Failure/return rates for all brands are below 5% for all manufacturers now. There was a dark period from 2011-2013 where a ton of terrible drivers and bad hardware shipped, but they're generally very reliable now. Everyone I know has moved to SSD for their primary drive, and are only using rotational drives for medium length local archival purposes.
moox. for a new generation.
I trust the name Toshiba. But I can't help but think any company aiming at the budged SSD market will skimp on wear leveling in favor of other attributes. Yes I saw the 7% overprovisioning note, but that was the whole of the attention given to a rather complex topic.
I wouldn't mind the somewhat slower access noted but in recommending an SSD for general system use I would be wary if a drive couldn't handle large volumes of throughput over its lifetime. Modern applications, not even Windows 8, are careful with disk usage. For example Windows 8 is happy to use an average 2% of drive access time for record-keeping alone on my own PC (I had to disable services to make that stop)
The average user is still running 15 junkware applications that tend to be written with any consideration but the user in mind. How much more important is it for them to minimize disk wear for an SSD? Since they're not, the'll need drives that can hack it. I saw numerous benchmarks in the fine article, but not the kind about which I care the most.
So I looked it up myself. Ten minutes of Googling found some information for other OCZ drives, but not this one. The only other thing I can add is this table of return rates for previous editions.
If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
In consumer grade SSDs and typical desktop workload, all it takes is bad luck. I've had SSDs failing in first 3 months of use and I have one which is still alive after 5 years of constant use. In enterprise use this really depends on the workload. If you get a lot of writes, then probably TLC is not a smart idea yet, so I agree that from some angles there are reliability and write limitation uses. But for these there are drives that can handle a lot of writes - you know, you can still buy enterprise class SLC SSDs - but these cost a ton of money.
Currently drives that outperform it, like the Samsung 850 Evo, match it on a cents-per-gig level.
This sort of forces one to ask the question, who does Toshiba think it's selling to?
Also, while people are touting Toshiba's "no hassle" warranty, my experience with Toshiba urges me to wait and see how much of a hassle it really is.
Chas - The one, the only.
THANK GOD!!!
Been using a Samsung EVO 840 as the system drive in my desktop for about 18 months now. Fast as hell, and no hint of trouble so far.
Il n'y a pas de Planet B.
I have only had one SSD fail on me, it was an early one. I have switched everything I have at work and at home to SSD (currently mostly Samsung 840/850, Crucial M500/MX100) and have never looked back, modern drives don't seem to have high failure rates and the speed difference is so great I would still use them even if they were unreliable and I had to back them up all the time.
Violence is the last refuge of the incompetent. Polar Scope Align for iOS
the write limitation is a non issue for 99.999% of the computers out there.
There shouldn't be any fucking write limitations, period. With a standard mechanical drive, the same sector can be overwritten repeatedly for the entire life of the drive itself. Yet, SSD drives are here to stay because the marketplace is full of suckers like you who are so eager to bend over for something "new and better".
First and second gen SSDs had much wider margins of error in flash. The 2 petabytes of writes are a result of that. I would expect you cannot get that kind of results with the new ones. I would be interested to see how these new consumer grade TLC SSDs handle non-consumer grade workload with a lot of writes. I think that it's good that the consumer grade drives are getting better and better, it's just that the drives that can handle some more workload than a usual laptop and still survive a few years are going to be more and more expensive. Does that Samsung warranty say that it's void if you exceed the write counter? YMMV?
They were solved years ago.
Just stick to Intel, Samsung, or Micron/Crucial, and avoid TLC.
Take your concerns up with the laws of physics. I'm sure they'll care about your opinion.
This drive is not the first to break the 40 cents/GB mark. OCZ's own ARC series is cheaper than these drives while performing better; Crucial's BX series is roughly the same price while performing much better. Around the 500GB mark, Samsung's 850 EVO is the cheapest and best performer.
The controller has Toshiba's name stamped on it, but is almost certainly a Phison S10. Furthermore the firmware has obvious problems with sustained writes.
First and second gen SSDs were garbage, people are reporting 2 petabyte write lifecycles on them.
I've got a first gen and a couple of second gen SSD's still slugging away. I swapped my first gen 60GB drive into my laptop a year or so ago after I put a ATADA SX900 in, I have to agree that there were some serious problems with 1st gen drives, probably the biggest problem I ran into was windows not playing nice and occasionally 'locking' the drive because of a bad write cycle. Sometimes it was recoverable, sometimes it wasn't, sometimes it was really bad and you had to get an unlock tool from the manufacture to fix it.
In terms of reliability now? Rock solid I'd say. The biggest downside to SSD's is similar to Bluray/DVD-R/W drives when they were new, and tossing out firmware updates every couple of weeks for xyz bugs. Now you don't hear about updates on the old write media drives at all, and it's becoming less common for firmware updates to be pushed out for SSD's now that they're getting very close to maturity as a technology.
I know someone will come along and say but xyz performance suxs compared to raid0, and let me remind that person that if you're using raid0 for speed you now have a higher failure rate then a SSD.
Om, nomnomnom...
Their website says 10 Years or 150TBW for the 256GB model and 10 Years or 300TBW for the 1TB model. TBW is "terabytes written". Which isn't the "2 petabytes to failure" marathon test that took 6 months to complete, but 0.3 petabytes written on a 1TB drive is still a lot and way beyond normal consumer usage. My unofficial opinion is that only about 128gb is "hot" and the rest of the storage on a 1TB drive is typically "cold". Even a professional video editor is going to have trouble topping out their warranty.
http://www.samsung.com/global/business/semiconductor/minisite/SSD/global/html/support/warranty.html
moox. for a new generation.
Yep. I'll happily take 10x the number of IOPS and limited writes (that are in most cases many years of regular service) over "infinite" writes and moving parts. I wouldn't want to keep spinning rust in service more than 3-5 years tops, and if all the SSDs I've used will survive this long, why should I use the slower solution?
Specialist Mac support for creative pros, Melbourne
Toshiba invented flash memory. I am hopeful that they do something decent with the OCZ brand.
"His name was James Damore."
And no, those machines are no all purpose computers, they are phone switches which just boot up, read their OS and configuration data from disk and then work solely from memory. Configuration changes which might cause a sector to fail after a write are seldom, but can still be handled by spare sectors on the disks.
SSDs aren't designed for being written to - anyone who says otherwise is a liar.
Thanks for the non attacking and informative answer.
I had controller failures on the "industrial" models we used. They just seem like a bad proposition right now. 0 failures on the spinning disk, but I will give it a few years.
I have determined that my sig is indeterminate.
This is my point - in an industrial environment, I can't afford a 3 month later failure. They may have improved things, but a spinning disk is a safer proposition.
I have determined that my sig is indeterminate.
When I read news like this I get excited and think I could buy X to improve my life, only to find out soon thereafter that X is not available where I currently live (Portugal) and when X comes to my place about a year later (if at all), I realize that it costs around twice as much as in the US and additionally requires VAT and delivery fees. :(
If you can figure out how to charge the gate capacitor without damaging the insulating layer, you'll be rich.
Until then, don't complain
I put a SSD in front of my spinning rust RAID as a cache and use writethrough or writearound. Even if the SSD fails I won't lose data.
thegodmovie.com - watch it
Yep, nothing at all wrong with doing that. Massive increase in IPOS by using SSD and dump data to the platters for long term and redundant storage.
Specialist Mac support for creative pros, Melbourne
Seriously price per gb isn't a barrier. They could sell the 120GBs for a dollar, it would just be money losing and stupid. But, it would smash the "barrier". When there ain't nothin' in your way, that ain't a barrier.
It is no longer uncommon to be uncommon.
Either last year or the year before that.
I'm using some as scratch disks - huge number of reads and writes. I'm ready to replace them when they die but they keep on going.
Yep. I did a cursory search on Amazon and found Samsung 850 EVO 250GB 2.5-Inch SATA III Internal SSD for $97.99, which is 39 cents per GB. This is just another MojoKid advert for his shit site.
Interesting? For a "works for me!" anecdote? Really mods, DaFuq? What is your experience with SSDs, 1? 2 maybe?
Well I have to have dealt with hundreds in the last couple years at the shop and its pretty damned obvious they still haven't fixed the driver controllers as they are still a complete and total CRAPSHOOT. Get a good one? Congrats you will get to write several times what the drive is rated for, you lucky dog! Get a bad one? Well I hope you didn't want that data because you will just flip the switch and the drive will be gone with zero warning, it won't even show up in BIOS/UEFI and the you will NEVER EVER get that data back!
Say what you will about spinning rust but I haven't had a single drive fail that didn't give SOME kind of warning before dying, be it delayed write failures, be it SMART, at least you got some kind of warning...SSDs? Fuck you, no warning. And has ANYBODY here had one actually fail into "read only" as the manufacturers use to tout? Anybody? Because I have yet to see that happen even once, instead the controller goes and your memories are gone FOR-EV-ER. Again at least with spinning rust you can grab another of the same model and build a clean box and do a platter swap, has anybody here actually been able to recover data from a failed SSD? I bet we won't find a single person.
So do I avoid SSDs? Nope have 'em in both my office box and my gamer box at home....as OS drives ONLY, why? Because I don't give a single fuck if I lose the OS, that's why! But all these people putting SSDs into laptops which they put all their pics and docs on? Well I just hope you have daily backups to the cloud, because the first time you flip that switch and find all your pics, including those of some now deceased relative you can never replace, is gone for good? Yeah I bet you won't be singing the praises of consumer SSDs anymore!
ACs don't waste your time replying, your posts are never seen by me.
FYI: There's a read performance bug on the EVO series drives. So if you haven't updated the firmware you should probably do it.
Om, nomnomnom...
The 850 EVO has been gettable for around 32 cents per GB for awhile: http://camelcamelcamel.com/Sam...
Ten machines here. Just one uses an SSD at all, and that only as a temp. drive for editing large audio files.
I simply don't need one for other purposes.
Machines here stay on. Not loading VMs or 60GB games. Apps used stay loaded.
And even on the machine with the SSD, I've only noticed a 3x speed improvement (3 month old Intel 530 series). Worth it to me, because I spend 10 hours a week editing, but otherwise nothing to write home about.
I say SSDs are over-hyped, and I'm an original speed freak, changing the refresh interval on my PC clone to gain, what was it, 2 or 3% performance improvement. Disk optimizing in the PC XT days before anyone else I knew. But SSDs? For the most part, just not that impressed.
I'd rather hear about Chrome finally fixing its "eating all your RAM" issue, because that truly turns your machine into a dog.
I come here for the love
Or, even better, Toshiba (and indeed no TLC).
What brand of ssd?
Greed is the root of all evil.
Oh, thought you said white people. My bad.
Also with the additional benefit of better system performance while rebuilding an array after replacing a failed magnetic drive.
Greed is the root of all evil.
If you need reliability and SSD performance, then you need to pay for enterprise grade SSDs. If money is too tight for that, then you can get some of the performance distributing the workload to more spindles - this increases IOPS nicely, though nowhere near SSD levels.
RAID arrays are a little tight in panel PCs. Speed was never a huge issue, it was a case of reliability in environments where we feared spinning disks would not do well.
I have determined that my sig is indeterminate.
ADATA, if I recall. I could dig up the failed ones somewhere.
I have determined that my sig is indeterminate.
No, the point is you probably chose the wrong SSDs. You either got budget SSDs, or got consumer grade SSDs which don't like their workload. For there to be a general problem with SSDs you'd first have to explain why so many people are having nothing but success with them, and frequently in hard-working situations.
Are drive I/O speeds important in your environment or are magnetic disks sufficient? If your use case does not require high drive I/O then any longevity issue (real or imagined) is moot and buying SSDs is a waste of money. If drive IO is important, then does your use case demand lots of writes? Does the OS support TRIM? Are you using a drive geared towards the numbers of writes your system generates? If so, then is is a simple calculation.. my application writes on average x bytes per y. SSD z can handle n write cycles. With that you can estimate your replacement rate and costs. SSDs are great, but they are not a panacea, and require proper planning to use correctly in many environments.
Silence is a state of mime.
Unfortunately it is not infinity writes or reads. Google grown defects. Magnetic platters fail from bit rot/failing sectors more frequently than head crash/mechanical failure in my experience.
Silence is a state of mime.
They have, don't buy cheap junk and look into over provisioning..
No sir I dont like it.
for the OS. For storage they still need time. I just bought another 4 TB Hitachi Deskstar for my tower because I need the room. I can't wait until the capacity and prices of SSDs match mechanical. I just wish they had something better than a half-truth telling SMART built in.
SSDs most often fail hard with no chance of data recovery.
I still recommend them to customers. With one caveat - I will not install one for you if you do not have a good (automated and preferably off-site) backup system.
Drive failure of any kind sucks ass (yes that is the technical term). If you are not heavily pushing backup solutions to your clients and warning them of them of the dangers in any drive system, the fault is yours. If you do inform and they balk at even If you do this, then the speed benefit from SSDs far outstrips any failures now that the price is reasonable.
Silence is a state of mime.
I did not choose the SSDs. They came with the industrial panel pc and were marketed as industrial grade.
I have determined that my sig is indeterminate.
Spinning disks work fine. We thought SSDs would be better because of the possibility the system could be subject to vibration. Performance was never a concern. The use case includes some fairly continuous logging, but it was not expected that log would rotate very quickly.
I have determined that my sig is indeterminate.
That sounds like a vendor issue using cheap SSD components, not a fault of SSDs in general.
Samsung 840 EVOs are certainly a nightmare, everyone one I deployed has needed be replaced. Samsung 830 Pro and 840 Pros I have deployed are still running. Enterprise SSDs are the real place you should be looking for reliable performance over time.
This is my sig. There are many like it, but this one is mine.
So you can overwrite the 256GB one 600 times, and the 1TB one 300 times. Why is the smaller one twice as reliable? Some mistake surely?
You haven't used any recent Seagates then :D. When I look to my right I see two spinners being recovered with ddrescue :(.
And it's been like that for the last 2 years. These large spinning consumer drives are just crap.
I have about 50% successful retirement rate in old hard disks. My drawer is full of them. It's not like they live forever.
Of course they did. They haven't made a stable product yet. Anyone who disagrees must not use their computer much. Anyone who moves lots of data around (video editors, Photoshopers, music makers, etc) Knows the OCZ name is absolute shite. I have been building PC's since the 7th grade 80386 FTW. I build custom machines for artists (Video production, DSLR Photography, Sound Studios) and work for a small company that is a 3rd party maintenance company for storage arrays. Toshiba isn't known for it's reliability either, I replace every toshiba drive in the office PC with a Hitachi or Seagate drive off our parts shelf. I bought OCZ Vertex, and OCZ Vector for different clients. They were also used here, and in my home PC. They all failed, and some when they came back "fixed" they still would only work for a week before failing again. This is keeping up with firmware updates. These SSDs are the WORST hands down. I have used brand X drives with more reliability. Now everything is Samsung PRO if the drive is getting used and not just holding mostly static data.
Define waste of money. Industrial panel PCs normally have a small fixed size storage requirements. Last time I checked any computer store the cheapest storage option was a SSD. The smallest HDD at 500GB cost more than the smallest SSD at 32GB. The same applies when scaled up to enterprise grade gear.
The 3TB seagates are terrible. For a panel pc, though we have had a lot of success with 2.5" 500Gb spinning disks (seagate mainly, but some western digital).
I have determined that my sig is indeterminate.
It's hardly a works for me. It's a works for most line. I'll take the hard *data* that the vast majority of people who have SSDs don't have catastrophic failures over an anecdote that one guy had to replace something that sounded like a batch of poor parts any day. The mods aren't modding up an anecdote, they are modding up their own experiences.
Interesting you mention drive controllers. There are several vendors who have not had any issue with drive controllers. Have you been buying nothing but Samsung and OCZ by any chance?
Now for an anecdote. I've had 5 spinning rust drives fail over the years. 3 failed without warning: 1 catastrophic in service, 1 never powered up after a shutdown, and one died when it was kicked over while running. Of the 2 other drives, one showed clear warnings from SMART data and the other just started silently corrupting data with no indication other than ZFS checksums failing. One of the drives which failed in service was a Seagate 7200.11, a drive with a failure rate that would give an OCZ employee of 5 years ago a hard-on.
With that in mind I have one question for you: Do you only use backups if you have an SSD? You are absolutely insane if you think one technology is magically more trustworthy than the other, or that one manufacturer can't suddenly overnight release an absolute turd of a product. WD's done it. Segate's done it. Matrox's done it. IBM's storage division basically went under because of it.
Yep, comparatively a real bastion of reliability that spinning rust.
Yes the dreaded ST3000DM001-1CH166. Terrible that one. We've had 17 out of 24 fail in a consumer NAS fail in less than 3 years. It's also a very widely deployed external drive since it's cheaper than WD and WD has its own external drive products.
If one of those selections was reliability, don't buy OCZ.
Let me guess you figured if you plugged them in a 220 volt slot you can double the speed.
Normally for laptops I have found SSD to be much more relabel, as they get moved and bumped around. Actually I haven't experienced any issues with SSD at work as well?
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
Surprisingly, no. We specified the SSD version of the PC (that runs on 24V incidentally) since we were concerned the motion of the door the device is mounted in might cause a spinning disk problems. Nobody ever turns off anything before messing with it. I suppose there is a possibility that they killed power while the damn thing was writing, come to think of it.
I would imagine, yes, if they failed a lot, you would find yourself having to re-label SSDs all the time.
I have determined that my sig is indeterminate.
For an enterprise environment there isn't much of an advantage of SSD.
1. You are using rather powerful servers: in general the 20/80 rule is effect 20% of the data is used 80% of the time. With Servers with hundreds of gigs of active RAM, most of the speed is going on in the RAM. With the buffered write, and the occasional lookup of data that isn't there. So you can have a slower drive because your drive isn't being used as much.
2. You should have a good RAID setup. RAID is for data protection and it also offers faster performance too.
3. Redundant servers they just don't sit idle, they do their share of the work too. splitting the usage.
Having SSD for most cases (not all mind you) on an enterprise environment isn't really worth it.
SSD are good for Laptops and mobile devices. So you can access data quickly, with less power. On systems with relatively low RAM.
Having an SSD on a Desktop PC, is kinda a crap-shoot, unless you get one of the really high performance ones, you probably won't notice much.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
I have exactly one of those disks left in a BSD box running ZFS. It is only a matter of time I suppose.... At least SMART gave me warning the others failed.
I have determined that my sig is indeterminate.
>Samsung 840 EVOs are certainly a nightmare
Since when. I've installed somewhere around 100 and have only had 2 DOA's and 0, as in zero in field failures over two years. I have even more Pros in the field without failure.
> I suppose there is a possibility that they killed power while the damn thing was writing, come to think of it.
Uh, yes. That would have killed them near 100% if you used old consumer SSDs. You should be able to find tests of this out there.
Most modern consumer and all enterprise SSDs will not have this issue.
Though in general controller quality definitely is still an issue (as well as stupid decisions like deciding that it is _correct_ behaviour to just disable the drive and not having it show even in BIOS once you run out of writes or basically anything going wrong _and_ not having a working tool to fix that and e.g. at least read out the data. To my knowledge none of the manufacturers seem even interested. I guess it's not their data after all, so why should they care).
Yeah...not so much. I've had good success using SSD drives as cache volumes, and I have seen significant performance improvements, even over RAIDs of spinning disks. Especially if said array was upgraded with an SSD cache. Don't even get me started on performance differences between ZFS RAID volumes with and without an SSD cache. And it doesn't take a big drive to be effective. 32-64 Gigabytes is all that's really needed. The most significant improvements I have seen so far is on database servers, where a great deal of processing is reliant on file I/O.
We sent out some panel pcs with SSDs in them, and so far we have had to replace them all with spinning disks... When is the reliability and write limitation issue going to be solved?
Just as soon as the idiots start looking at reliability reports instead of benchmark speeds
(ie. never).
No sig today...
He's right, though. Throughout the era of SSDs there has been concerns about Flash cell longevity. The medium of HDD does not degrade meaningfully.
When is the reliability and write limitation issue going to be solved?
Perhaps you chose the wrong SSD's?
I was skeptical of SSD's - then I tried a computer with one.
Yeah I know, sample size of one.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
He's right, though. Throughout the era of SSDs there has been concerns about Flash cell longevity. The medium of HDD does not degrade meaningfully.
And yet, hard drives can and do crash. They fail. I've replaced dozens over the years. What good is "no write limitation" on a hard drive when the head augers into the disk? When the platter bearings go south? When the media starts flaking off?
To me, the write limitation is moot. It's just a different failure mode. I haven't had enough experience with SSD's to make a definitive judgement, but I do like a lot about them so far.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
Perhaps, but it is not necessarily the case that I can gamble with tech that has let me down.
I have determined that my sig is indeterminate.
I have some old systems around (some of them running since more than 20 years) with old 2- and 4-GB-SCSI disks. While they read fine, you should not try to write onto them.
There's no reason not to try to write those, because SCSI disks have intelligent controllers. You shouldn't try to write your antique ST-506 interface drives, but how many of us are even in a position to hook any of those up? I guess I still have a PC with ISA slots, it's a Geode LX dev board...
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
It's easy to spot a SSD fanboi. They always use the term "spinning rust." You used it twice, in fact.
It's o.k., you don't have to be polite here....this is Slashdot! Don't hold back.....tell us how you really feel!!
So I can quit seeing the phrase "spinning rust" on Slashdot all the freaking time.
I've also had HDD's failing in first 3 months of use, many more than SSD's. SLC SSD or RAM SSD's are necessary though to maintain throughput on modern systems. No way spinning rust is keeping up with a pair of SSD's, heck even USB Flash drives perform better as a boot drive than HDD's. HDD's only make sense if you need really large amounts of storage for cheap and even then you're still inserting SSD's and RAM as caches and accelerators.
Custom electronics and digital signage for your business: www.evcircuits.com
TRIM is pointless as you need a consistent output/throughput in the worst case scenario Also, in most cases, your workload will be optimized against the hardware so you won't be doing any such small writes/updates where TRIM becomes useful (stripe size on an FS that requires frequent updates is already 4k or larger).
Most people simply do not know whether or not IOPS are important but they generally are. Given that spinning drives peak out at ~100IOPS, SSD's are generally a better investment (also in power and space requirements) than multiple drives (or drive arrays). You also don't try to run any system without at least having RAID1 and regular snapshots/backups, that way, your SSD's can fail every month, you just ship them back to manufacturer.
SSD's are never a waste of money IMHO unless they are literally unused. They can always be used for something, heck I have paid for the very first Intel 32GB SLC SSD's. Still being re-used in a small backup server systems.
Custom electronics and digital signage for your business: www.evcircuits.com
If you're putting commodity hardware in an industrial environment, the fault lies with you, not the hardware. There are many issues that could contribute to device failure in general: RFI, power line harmonics, localized overheating, dust, electrically conductive dust, ... Not to mention buying brands that are notorious for their shoddy quality.
So what good is $0.40/GB if the thing fails in less then year. OCZ has been a price leader for years, also one of the biggest reliability failures. I would not use their stuff if it were free. If you want an SSD with reliability buy a top tier brand like Samsung or Intel. Sandisk is probably decent, but I haven't used theirs.
All HDD and SSD will fail eventually, but OCZ had the worst reputation of all SSD makers. If your data is lost or system crashes do you really care how cheap the drive was?
"Panel PC"
That's your problem. Heat still kills electronics. Learn proper thermal management and this wouldn't be a problem.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
Everyone who cares about their data should have redundancy. Arguing about quality is stupid and reliability of a single drive if the halflife of the drive is measured in years. Why switch to something that has 50% longer MTBF when you can have two redundant units and get 50000% longer MTBF?
Disks are slow, and SSDs are a small price to pay for fractional random access latency. There is simply no comparison. All computers should run an SSD unless they have specialized requirements that prioritize size and large read/writes -- video perhaps. If the data is valuable, there should be two drives running RAID1 regardless of whether they are SSDs or HDDs.
Toshiba is garbage. Overheating laptops and failing hard drives are the bulk of my experience with their computer products.
Perhaps, but it is not necessarily the case that I can gamble with tech that has let me down.
Platter hard drives are a very common failure item. I've replaced a lot of augers and singers in my day.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
And I have 6 SSDs currently, none of which have failed. On the other hand, I also have 13 rust buckets (working), and I'm approaching a graveyard of about 11, all of which are less than 3 years old. Some of which just instantly died.. No SMART, no warning, just dead. Some died during a power cycle and didn't come back. Average time to death for the rust buckets is about 1 year. Some drives have been replaced twice already (The replacement was replaced). The SSDs are about 4 years old on average. I'd have seagate repair them, but the cost of shipping them to seagate and back, and then getting a drive I know will die again shortly just isn't worth the cost. Just replacing them with another model as they die.
All drives are kept cool, on a top tier power supply which is connected to a UPS. So it's neither a power problem, nor a cooling problem. And the replacement rust buckets aren't dying. So your theory of how awesome and bulletbroof rust buckets are, is just a anecdote is they are susceptible to crappy failure rates and instant death syndrome that you describe.
I'll take the SSD write "problem" any day. Rust buckets just like anything else are a crap shoot. I bought the wrong model from the wrong manufacturer. Oddly, the rust buckets are mainly write-once, read seldom, while the SSDs get pounded consistently yet it's the rust buckets that are dying.
I remember buying ssd's in 2004 for some of our field machines whose early 90's hard rives had failed (possibly related to a power surge / lightining strike at the customers site). Being a install and forget (until it breaks) system I found some ssd's with 128Mb for 300 GBP they came with an 8 year guarantee.
Being a DOS based system these things booted from post to prompt in a blink of an eye, 486 industrial pc's over ide, the latency figure is what made all the difference, not the throughput.
I still don't get why ide ssd's never took off enough to become cheap, they wouldn't need to be large. Those early 2000 laptops covered in dust would fly compared to the spinning discs in them. (I have even heard of people using the sata to ide adapters to get a ssd into an old laptop, some hacksawing required to make room in the chassis though I am dubious of the success rate with old bios's etc ). When this is brought up on forums people often say that the ata bus won't see an improvement, however the latency saving makes a dramatic improvement even if the throughput doesn't increase at all. But now XP is dead and windows 8 requires the exec bit in the processor the old processors can't be used (unless someone is willing to correct me) so unless the user is willing to accept linux there is little point in reviving them for desktop use.
I chose the crucial 128gb m4 in 2009 for my pc because the reliability indicators were so much higher than other cheaper faster bigger drives. And it is still ticking along (though I made sure swap was turned off and tmp was on a hard disk).
Yeah, that'll be the problem.
"industrial grade" is typically code for "industrial temp" and they just pick whatever crap meets that requirement. And knowing that you probably want that, the controllers are probably full of bugs.
I'd actually trust the good consumer version of the SSDs than the buggy industrial ones where the technology dates to before Intel SSDs made them good.
We've improved things a lot -SSD failure on power loss is generally limited to the cheap non-name brand crap or they have built in power protection, speeds have exceeded what SATA can do, etc.
Heck, sometimes what's inside those "industrial ssds" is probably just a CompactFlash type card which have poor speeds and so-so reliability.
By far the biggest failure among dozens of SSDs deployed is power failure induced loss, and it was rare. We did limit ourselves to good quality SSDs from Intel and Samsung. The data loss happened when the owner dropped his laptop while it was on and it shattered into many pieces. (Laptop was a write-off, as well). But the SSD was recovered by a secure erase.
In an industrial environment, I can see sudden power loss crapping out the SSDs - no one takes the time to properly shut down the PC, and industrial SSD vendors typically don't follow the advancements in SSD technology that's happened in the past few years.
I have/had a 11-drive NAS using those drives. I have 2 drives left. 2 were replaced twice. So I started with 11, have 11 dead, with 2 still working. That's some bad stuff.
And yet, hard drives can and do crash. They fail.
Three rules of thumb:
A) Avoid heat
B) Avoid vibration
C) Avoid Seagate!
Go google evo 840 write amplification bug? Within 90 days they slow down to a halt without a firmware update
http://saveie6.com/
Interesting? For a "works for me!" anecdote? Really mods, DaFuq? What is your experience with SSDs, 1? 2 maybe?
Well I have to have dealt with hundreds in the last couple years at the shop and its pretty damned obvious they still haven't fixed the driver controllers as they are still a complete and total CRAPSHOOT.
We have about 1100 of several brands and types deployed in our organization, and have seen a lower than hard disk failure rate, even on the cheap OEM Lite-On drives. How's that for an anecdote? (I do agree about the controller problems - we had to push out a firmware update for those Lite-Ons to fix a Deep Sleep issue.)
2-3 years ago there were "growing pains" in the industry as everyone was new to it. I lost several drives in that time.
All of my drives of the last 2 years still work perfectly.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
a spinning disk is a safer proposition
Tell that to my Seagate drives...
Peter predicted that you would "deliberately forget" creation 2000 years ago...
My 840 is working great on the latest firmware.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
Are there different versions of the ST3000DM001? I guess so, cause I'm seeing several listings of, apparently, that very same model, on the same shop as different "models", with slightly different prices. (I hope I explained myself here)
I also wonder if the newer ones are any more reliable.
Thanks
I haven't had a single drive fail that didn't give SOME kind of warning before dying
Apparently you're just not old enough.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
C) Avoid Seagate!
Amen!
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
I wonder if it is possible to start a class action lawsuit ?
SSD cannot displace the nearline functionality of hard disk until it gets within a factor of two in price. BTW, nearline is still expanding exponentially with no end in sight.
Using myself as a predictive example... My workstations all have spinning disks in them, and each has at least one SSD for booting and serious work. The hds are normally spun down, which does wonders for noise and lifetime. The ssds are normally 90% full.
When all you have is a hammer, every problem starts to look like a thumb.
Silicon power SSDs, which my shop has sold over 300 of, have had a failure rate of zero drives thus far, they're medium-high speed compared to other drives, and they're $0.35 per GB. They're still the king in best overall SSD by a long shot.
Yay, the media can be written to repeatedly! Well, except when the the RW head assembly reaches end of life after one too many head parks/unparks.
For an enterprise environment there isn't much of an advantage of SSD. 1. You are using rather powerful servers: in general the 20/80 rule is effect 20% of the data is used 80% of the time. With Servers with hundreds of gigs of active RAM, most of the speed is going on in the RAM. With the buffered write, and the occasional lookup of data that isn't there. So you can have a slower drive because your drive isn't being used as much.
Two points here:
a) RAM is still way to expensive to manage the 20% of the data. Put the hottest 1% in RAM, 20% on an SSD and the rest on HDD
b) if your workload changes, then you start getting a lot of cache misses your performance is going to be slaughtered.
2. You should have a good RAID setup. RAID is for data protection and it also offers faster performance too.
couldn't agree more
3. Redundant servers they just don't sit idle, they do their share of the work too. splitting the usage.
This is not the point here, this is true regardless if you use SSDs or not, as you can scale out both. And redundant servers in a workload balancing setup are only acceptable if you can afford performance drop when one of the redundancy group elements fails. In some extreme cases in such setup the degraded system will not cope with full workload and will fail.
Having SSD for most cases (not all mind you) on an enterprise environment isn't really worth it.
There is a lot of enterprise workloads that require SSDs. Just look at sales figures.
SSD are good for Laptops and mobile devices. So you can access data quickly, with less power. On systems with relatively low RAM.
SSDs are good everywhere where the users can feel the performance benefit. My current desktop does not have an SSD, because I did not feel the need, but if I needed to buy a new one I would definitely use an SSD for a system drive.
Having an SSD on a Desktop PC, is kinda a crap-shoot, unless you get one of the really high performance ones, you probably won't notice much.
Assuming you're not having a 15k rpm drive in your desktop, the difference is definitely noticeable.
Go back about 10 years and it was avoid Western Digital.
Sorry, teleporters just kill you and then make a copy. A perfect, soul-less copy.
>Having an SSD on a Desktop PC, is kinda a crap-shoot, unless you get one of the really high performance ones, you probably won't notice much. I beg to differ. It used to be the upgrade that gave you the greatest performance increase was increasing your memory. Now a days going from spinning rust to ssd provides the most noticeable performance upgrade.
Sorry, teleporters just kill you and then make a copy. A perfect, soul-less copy.
The OCZ Trion 100 is lagging behind the competition in terms of performance and price, so why such a big deal? See this review by Anandtech (http://anandtech.com/show/9408/ocz-trion-100-240gb-480gb-960gb-review/11).
Right but these in particular, OCZ is kinda renowned for being unreliable across their entire product range.
I remember old Connors sucking balls... and getting an *entire batch* of Fujitsus that were bad (eighteen years ago)... IBM "Deathstars" were virtually guaranteed to go belly-up soon after purchase (fifteen years ago)... but I'm not aware of Western Digitals from a decade ago being likely to have problems disproportionate to their numbers... in fact, I've low-level-wiped and run diag on *so very many* WD's from that time period (easily in the thousands) that I feel sure I would have spotted a trend...
Yeah, and the first ones to avoid were the CMIs in the early 80s. They actually dumped an artificial reef of them in the ocean, they were so bad. Absolutely true you can't just apply statistics forever. But for the present, Seagates are absolute unmitigated disaster garbage, and WD isn't all that much better. HGST (Hitachi) is the only worthwhile one left in company-wide terms (knock on wood for the future); the GOOD Toshibas; the ones whose design came from HGST; are fine too. But there is no non-anecdotal way remaining to tell them part. I just know that my big bunch of Toshiba DT01ACA300 3TB's are definitely HGST's design. I know that because some of them even have HGST on the l;abels, and all of them sport identical firmware, no matter what the labels say.
HGST is only "sort of" independent. They actually got bought out by WD, but it is a wholly-owned subsidiary which so far has operated completely independently and not succumbed to the WD malaise. Hitachi the mother company now owns 10% of WD and designates two of the board seats, believe it or not. Wheels within wheels.
Since SSD's have been known to have catastrophic failures why not market drives that for all intents and purposes can't fail or can easily be repaired by changing fuses or other simple components? If it's the controller that is failing why not have a second low performance backup controller that only works in read only mode? It just doesn't make sense to me that they can't make these things 100% read only reliable or that bad parts other than the flash can't be relapsed.
I suspect that, if you look carefully, you will find that there was indeed a mistake. It may not be where you are expecting.
"So long and thanks for all the fish."
I know it's not infinite. Defects, bit rot, failure of moving parts - there's a lot to go wrong in a hard drive, that's why I had "infinite" in inverted commas...
Specialist Mac support for creative pros, Melbourne
Ocz are garbage, wouldnt use their gear if it was given. Theyve had problems with nearly everything theyve made. Utter shit.
My guess: either it's a mistake, or the flash controller has access to significantly more memory cells on the 256GB drive than the outwardly visible 256GB because they're using chips that wouldn't pass muster (in terms of good cells to bad) for the 500 GB / 1 TB models, and so are then able to spread cell wear around more effectively. This has been common practice by CPU manufactures for a while (de-rating a CPU that wouldn't run right at xxx Hz, but will run fine at yyy Hz)
Actually I was using the term the parent used when replying to the parent. in FACT!
Right but these in particular, OCZ is kinda renowned for being unreliable across their entire product range.
Was. OCZ *was* renowned. OCZ was once it's own company. OCZ exists now in name only and any comparisons to the times of old make no sense anymore.
That said (touch wood) I still have a working OCZ Vertex 3. Now there was a drive with a horrendous failure rate.
It's only a crap-shoot if you buy shit brands. Stick to the names that are known for memory - SSDs are just Flash memory, after all. Crucial, for example, knows what the heck they're doing. Their Storage Manager tool even supports UEFI reboot-to-ISO-image for firmware updates.
I'm guessing that yes older machines may"fly" to an extent, but the full capacity of the through put will never actually be realized do to bus speed of the architecture of those machines
I have an old SeaGate 478MB drive that was working last time a checked at least.
Perhaps it would've been safe to assume that my comment was intended to be at least partially useful; i.e. I wasn't referring to drives that are so fucking old that they're of legal drinking age?? ;)
SSD reliability issues include in rough order of importance:
1. Corruption on Power Loss
2. Trim Corruption
3. Unpowered Retention
4. Write Endurance
The problems with trim are annoying but no SSD should need to use trim for good performance. Those that do were designed poorly:
http://www.realworldtech.com/f...
Corruption on power loss in this case is not corruption of data being written at the time of power loss; that is expected. It is corruption of unrelated data or state which may render the drive unusable. Some SSDs do not suffer from this problem and it should not occur in a portable application because of battery operation but a desktop is different matter.
I have used Compact Flash to IDE adapters to replace old hard drives however apparently I have also been lucking in my choice of Compact Flash cards; like almost all Flash storage except for some SSDs, most are terribly unreliable.
Transcend among others makes IDE Flash drives intended for industrial applications:
http://www.transcend-info.com/...
How was that a troll?
Real world experience, that goes against the hive mind, is a troll?!
If so, what is the point in commenting on anything?
Either one agrees with the hive, or one should shut up, apparently.
Personally, I am less worried about Slashdot's self-destructive "beta" and more worried about the group non-think. But in saying that, I must be making another troll remark, right?