What do the words solid-state storage mean to you? A whole bunch of attributes come firmly attached to the technology. They might include expensive (when used in SSDs), fast, robust, cool (physically and arguably culturally), low-power, silent, convenient (such as when packaged into USB sticks, SD cards and the like) and reliable.
And when you think about their application in enterprises, you might lengthen your list of characteristics to include performance-enhancing (ahem).
But although we know that that SSDs do wear out, we also think that this problem is a known one. In most SSDs, controllers check the condition of the flash chips and level out the wear so as to lengthen the overall life of the device. They are often over-provisioned to allow for chip failure while remaining inside the stated capacity. Failures are therefore likely to be gradual.
And when I spoke recently to Texas Memory Systems, one of a growing band of hardware vendors who sell just SSD-based storage, about their new 12TB SSD appliance, they said that everything in the appliance is doubled up. "There's no single point of failure," TMS senior analyst Erik Eyberg said. He said there are four layers of error correction, redundant everything, and proprietary RAID technology that allows the data to be rearranged to handle the failure of flash chips efficiently.
So the SSD problem is solved -- especially if you can afford to pay the $240k that TMS is asking for its 12TB storage array because you really really need that SSD performance.
Are SSDs reliable enough?
But is it solved for the rest of us? Come a little closer to home and I can tell you that an SSD attached to my PC failed this week. Before you shed tears and to pre-empt the build-up of any unnecessary tension which would be detrimental to your blood pressure, let me reassure you that no critical data was lost.
My main PC houses a 128GB SSD that boots into Windows and contains just one or two critical applications, a second 128GB SSD containing programs, and a spinning disk that holds my data. The boot and data disks are backed up, the applications SSD, made by Patriot Torqx, was not -- though that turned out not to be critical.
You don't expect SSDs to fail and when they do, you expect them to give some sort of warning. This one, only 18 months old and according to SSDLife only 61 percent worn, didn't: on Tuesday morning it just refused to talk to the PC or any other machine I plugged it into.
Luckily, because Windows retains application settings in a separate location, all I lost was the program code. I replaced the SSD with another one and reinstalled the applications, which was a little tedious but fairly painless -- even QuickBooks, which has about four levels of anti-theft protection, each more annoying than the last.
Maybe it was just a rogue device. Maybe it was a batch problem for Patriot Torqx, which has a lot of users on its forums and elsewhere complaining about lost data after sudden failures. It certainly was a standalone device without the benefit of redundant everything. And something I hadn't fully appreciated is that, when an SSD fails, there seems to be little chance of recovering data. If the saga develops and I get the data back or even a replacement unit, I'll report back.
So think a little further beyond the benefits of SSDs when using them in a desktop. Stand-alone devices don't deliver all the benefits of high-end or enterprise-level gear and failures can be both sudden and catastrophic. Physically robust they certainly are but failure-proof? No way.
Twitter 
Linkedin 
ZDNet UK App 
Facebook 
RSS Feeds 
Newsletters 
Talkback
Very interesting, and thought-provoking. You make a good point that I hadn't thought about, when a spinning disk "fails" I am almost always able to get some/most/all of the data from it. If an SSD device failure generally results in immediate inaccessibility of all data, that is a very significant difference.
jw
I think until the price of these SSD devices becomes more accessible, people will try to work around the cost with cheaper devices which don't contain the necessary redundancy, which is ok, until something goes wrong..... I think I would prefer to wait for the technology to improve and the costs lowered so I can afford a more robust device.
I've tweeted about the scary SSD failure rate. See Jeff Atwood at Coding Horror
http://www.codinghorror.com/blog/2011/05/the-hot-crazy-solid-state-drive-scale.html
He writes:
Portman Wills, friend of the company and generally awesome guy, has a far scarier tale to tell. He got infected with the SSD religion based on my original 2009 blog post, and he went all in. He purchased eight SSDs over the last two years … and all of them failed. The tale of the tape is frankly a little terrifying:
•Super Talent 32 GB SSD, failed after 137 days
•OCZ Vertex 1 250 GB SSD, failed after 512 days
•G.Skill 64 GB SSD, failed after 251 days
•G.Skill 64 GB SSD, failed after 276 days
•Crucial 64 GB SSD, failed after 350 days
•OCZ Agility 60 GB SSD, failed after 72 days
•Intel X25-M 80 GB SSD, failed after 15 days
•Intel X25-M 80 GB SSD, failed after 206 days
Maybe certain kinds of SSD are more prone to failure than others. I have owned an Asus Eee 701SD netbook with an 8GB SSD, since October 2009, and it has shown no sign of problems (though from what I'm reading here, failures can be sudden, silent and deadly...).
I don't know if/how it's possible to do this in Windows, but in Linux there are ways to minimise the read/write cycles (and therefore wear) on SSDs as much as possible. For instance, on my Eee running Arch Linux, I have no swap partition (though it helps that the machine has 2GB RAM installed), and there's a temp directory for short-term log files, etc. which is set up as a ramdisk.
@Jack: I don't know whether Portman Wills was writing heavily to those SSDs, but those failure times are highly alarming. I put the important files on my Eee into the Dropbox folder, but if the Eee's SSD failed totally all of a sudden, I'd have a fun job piecing the OS setup back together again...
@Jack: that really is scary. I'll work on my backup routines...
@Tim Walker: Windows 7 does understand SSDs - it refuses to defrag them which is good - but whether it goes so far as to adjust its writing strategy, I'll need to do some more research.
I have been trying to recover important and almost unique data from a friend's failed 4Gb Quantum USB memory stick with absolutely zero success despite using commercial tools suitable for recovering lost partitions and lost data etc.
I guess there's a a strong parallel between an SSD hard drive and a USB memory stick.
Independently, I have been reading about SSDs and had already decided against trying one, mainly for the speed.
A rigorous backup regime would be absolutely essential when using an SSD, something many people do not do.
Something else rather worrying that I just thought of: the use-case here is a desktop PC, with one or more SSDs used for quick-booting the OS, and with a spinning disk for the rest of the data. What about laptops like the SSD-equipped MacBook Air models, where the SSD is the only internal storage device installed?
If I ever owned an SSD Air, I'd make certain I was backing up to a Time Machine host (and probably Dropbox and/or Wuala) at every opportunity. Not to mention that I'd be pretty honked off if my best-part-of-a-grand dream machine lost its "brain" after only a few months...
The big push with SSD initially was the fact that they don't have moving parts. But after seeing this it makes you wonder which is better, spinning disks or SSD. I suspect a lot of businesses will be staying with spinning disks for now mainly for cost reasons, especially since redundancy is needed with whatever drives are chosen anyway.
@Tim: agree wholeheartedly. My laptop has an SSD (which is nice for performance and boosts battery life of course) but the only place I put data is in the Dropbox folder, which also syncs to the desktop. I don't trust it...
Great post! :) Maybe the device mentioned above is not fully understood by other people.. So therefore needs more information regarding SSD.
>> Failures are therefore likely to be gradual.
umm, i think not. see below.
this is the lifespan of all the ssd's ive installed :
ocz solid - 47hrs
ocz vertex - 3 months
ocz agility - 11 months
compare that to mechanical drives ive installed, about 40 over the last 15 years, and only one developed a corrupt sector that warranted backing all my precious data up then reformat to fix. when ssd's fail, they fail bigtime with no warning and you are left with a brick. files - gone, emails - gone, windows - gone - all in a flash.
On March 23, 2012, I built a new system. Antec 100 window box, MSI 723 mobo, AMD FX 8 core processor, 600 watt power supply, 8 gig DDR3, 128 gig Samsung SSD. Win 7, 64 bit, and Linux Mint, 64 bit. Speed is astonishing. One major problem though, win 7 will only run for about 2 minutes before shutting down. Still trying to find a solution. Mint has no problems, and is considerably faster than win 7. I have a 500 gig regular HD attached also for backup. Anyone had similar problems with win 7?