Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Data Storage

Reliability of PC Flash SSDs? 467

An anonymous reader writes "SATA and IDE flash solid-state disks are all the rage these days — faster and, allegedly, more reliable than traditional spinning-rust disks. My organization dipped its toe in the flash-disk waters, buying a handful for some PC and Linux boxes. Out of 8 drives from various manufacturers, 3 have failed in the space of four months! Some are reporting bad blocks, others just crapped out and stopped responding entirely. (And no, this isn't a wear-leveling issue, nor were these machines in particularly harsh environmental conditions, nor were all failed drives from the same manufacturer.) So I ask you, the readers of Slashdot: what has your experience been like with basic, consumer-grade SATA or IDE flash drives? Are they failing for you too, or are we just unlucky? It's starting to remind me of the claims about long-lifetime compact fluorescent light bulbs that, in reality, have turned out to be BS!"
This discussion has been archived. No new comments can be posted.

Reliability of PC Flash SSDs?

Comments Filter:
  • Don't Defrag (Score:4, Informative)

    by Anonymous Coward on Tuesday October 27, 2009 @12:20PM (#29885041)

    Make sure you turn of the scheduler for defragging in Windows or whatever OS you are using. Defragging those types of drives will effectively kill them.

  • by Anonymous Coward on Tuesday October 27, 2009 @12:25PM (#29885093)

    Hi, I was the guy that posted the original question. Thanks for your response. I didn't give details simply due to space constraints. The drives were:

    1. FHM16GF25H = Super Talent MasterDrive 16GB under linux
    2. Transcend TS32GSSD25-M under Windows/XP
    3. Patriot Warp v2 32GB under Ubuntu 8.04 with ext3

    The machines were not super heavily loaded (i.e., no compiles 24/7), and we did the "obvious" things like turning off atime updates to the filesystems, etc.

  • by initdeep ( 1073290 ) on Tuesday October 27, 2009 @12:26PM (#29885109)

    you mean the real world support for TRIM in Windows 7 and supported in Indilinx and Intel controllers?

    the one that has been recently tested out on Anandtech and shown to have very positive results?

    oh yeah, that one.

  • Re:Don't Defrag (Score:5, Informative)

    by golfbum ( 1408137 ) on Tuesday October 27, 2009 @12:28PM (#29885147)
    defrag benefits hdd due to their long latency to retrieve widely separated block of info. ssds have essentially no latency therefore don't benefit by such reorganization. gb
  • Re:Don't Defrag (Score:5, Informative)

    by Reece400 ( 584378 ) <Reece400@hotmail.com> on Tuesday October 27, 2009 @12:31PM (#29885195)
    Lots and lots of extra reads and writes, which are unnecessary as SSDs do not benefit from defragmentation.
  • Re:Don't Defrag (Score:2, Informative)

    by gregthebunny ( 1502041 ) on Tuesday October 27, 2009 @12:33PM (#29885209) Journal
    Wow... Google much [google.com]? http://www.tomshardware.com/forum/246283-32-defrag [tomshardware.com]
  • by Anonymous Coward on Tuesday October 27, 2009 @12:34PM (#29885231)

    Hi, I was the guy that posted the original question. Thanks for your response. I didn't give details simply due to space constraints. The drives were:

    1. FHM16GF25H = Super Talent MasterDrive 16GB under linux
    2. Transcend TS32GSSD25-M under Windows/XP
    3. Patriot Warp v2 32GB under Ubuntu 8.04 with ext3

    The machines were not super heavily loaded (i.e., no compiles 24/7), and we did the "obvious" things like turning off atime updates to the filesystems, etc.

  • by initdeep ( 1073290 ) on Tuesday October 27, 2009 @12:43PM (#29885353)

    I'd be more looking at the fact that all of those are JMicron based controller drives and are shitty examples of SSD's in the first place.
    http://www.anandtech.com/storage/showdoc.aspx?i=3531&p=17 [anandtech.com]

  • by justthinkit ( 954982 ) <floyd@just-think-it.com> on Tuesday October 27, 2009 @12:54PM (#29885521) Homepage Journal
    Linus updated his SSD post [blogspot.com] 5 months later and in the follow-up mentioned, among other things, an AnandTech article [anandtech.com] he liked at least parts of.
  • by commodore64_love ( 1445365 ) on Tuesday October 27, 2009 @12:56PM (#29885539) Journal

    We've not being seeing widespread failure of Ipods or other keydrives, even though they use the same F-RAM technology. I'm kinda surprised to hear any reports of failure in the new solid state PC drives, unless it's an issue of making the cells too small to be reliable.

    Aside - I have two traditional hard drives in my PC. They've been spinning almost-nonstop since 2003. Any idea how much longer I have until they crash?

    Aside #2 from the Summary -

    - The savings on CFLs is trivial. I might switch my bulb from 40 to 10 watts, but I still have a 10,000 watt heat pump running. I'm not seeing smaller monthly bills.

    - CFLs hate temperature extremes. CFLs hate dimmers. In practical terms this means CFLs can not be used in 80-90% of present fixtures, like those that are enclosed (heat kills CFL electronics) or outside (too cold to light) And I bought a so-called "dimmable CFL" which died 5 minutes after I installed it in my living room dimmer switch.

    - CFLs hate being turned on and off. Rapid cycling makes them die even faster than an incandescent bulb (as stated in the summary). So you've spent 5 times as much for a bulb than doesn't last any longer.

    - CFLs have a warm-up time. The 60-watt-equivalent bulb hanging upside-down in my kitchen is sometimes so dim, it looks like a brown dwarf star... barely any light at all. It takes 3-4 minutes to finally reach full brightness.

  • by tolan-b ( 230077 ) on Tuesday October 27, 2009 @12:57PM (#29885557)

    There's no public trim on the Intel controllers yet.

    Do you mean the experimental trim support in the beta Indilinx firmwares that caused data corruption when your computer went into sleep? Great! Those drivers got pulled for obvious reasons.

    The offline 'trim' doesn't count btw, it's not using the trim command and you have to run it manually periodically rather than it running automatically when the disk's idle.

    Trim will be great but don't pretend it's available.

  • Re:Don't Defrag (Score:5, Informative)

    by Zerth ( 26112 ) on Tuesday October 27, 2009 @12:58PM (#29885579)

    Is there a lifespan advantage to be had from moving all your files around the SSD once in a while?

    eg. You could move the least-used cells to the most-used cells to even out the wear

    Any {dr}ecent controller does wear leveling

  • Re:Don't Defrag (Score:4, Informative)

    by TheRaven64 ( 641858 ) on Tuesday October 27, 2009 @01:04PM (#29885651) Journal
    There is, but that's why the controller does it for you. It does this based on how long it's been since a given block was written to, and it tries to consolidate infrequently-written blocks into the same cell. Running defrag messes up this heuristic.
  • by Concern ( 819622 ) * on Tuesday October 27, 2009 @01:04PM (#29885657) Journal

    Thank you. The brands/models were the critical piece of information.

    You're probably aware that SSD's have been in the server space, at a very different price point, for a few years now, without any extraordinary reliability debacles. To some extent, this is a case of getting what you pay for. I did a moderate amount of research on SSD drives, relying especially on the independent review sites, and quickly eliminated all of the brands you described.

    As is frequent in fairly new markets, there are a few smaller and less well-run companies trying to dive in, and their first customers get to beta test their v0.* and v1.* offerings.

    The prevailing wisdom seemed to me (and to people like i.e. Torvalds) that Intel was far and away the top of the heap in terms of performance and reliability, and some drives based on a newer Samsung controller (i.e. OCZ Summit) were a perhaps credible alternative. Other brands were clearly struggling to even be in the game, with frequent firmware updates and outright debacles (i.e. Indilinux, Micron) and we're in the process of shaking out who will make it and who will not.

    I have only fielded a few consumer-grade SSDs over about the same amount of time as you, but going with Intel's G1 and G2 MLC products has so far yielded zero failures.

    If you are already in the market for an SSD, and you are ready to spend premium money for premium performance, you should go the whole distance and go with Intel, the current market leader. See also the latest news on these models. [tomshardware.com]

  • by jggimi ( 1279324 ) on Tuesday October 27, 2009 @01:07PM (#29885683)

    Any idea how much longer I have until they crash?

    While nothing is ever a certaintly -- a tool for your OS that inspects SMART data from your drives' electronics would answer that question, at least from a trend perspective. I like smartmontools [sourceforge.net], but you may prefer something else, or it may not be applicable for your OS.

    See Wikipedia [wikipedia.org] for some background information on SMART, and what it can tell you.

  • by Anonymous Coward on Tuesday October 27, 2009 @01:09PM (#29885733)

    Leaving a CFL turned on for less than 15 minutes at a time lowers the life expectancy. It's something to do with the inverter not warming up completely, I don't remember exactly.
    I had several CFLs die within 1-2 years, as opposed to their advertised long life. I did some research and then changed my usage patterns. I installed an incandescent in the bathroom for the quick in-and-out. In the rest of the rooms, I will leave the light on when I leave, then turn it off when I have re-entered and exited the room again. Since making these changes over 2 years ago, none of the CFLs have died.

  • by valhallaprime ( 749304 ) on Tuesday October 27, 2009 @01:10PM (#29885737)

    It really depends on the brand of the bulb. I've had a few Philips that were bought in the 90's, used every night dusk till dawn outside, and they lasted 10+ years. In our school dorms, I replaced a few spots in the common area with a few of the older looped (not squared-off) Ikea 11W'ers. Light's are on 24/7/365. 6 of the 8 are at 2+ years now, that's almost 20,000 hours, on -already used- bulbs.

    We replaced all the hall lights in the dorms with 13 watt and 20 watt CFL's, for a total of about 45 bulbs. All GE brand....4 have failed after 18 months of 24/7/365. The rest are still going strong. That's still way above their spec of 8,000hrs IIRC.

    I've used a few FEIT and Lights Across America. One LAA had a "bad failure", where the ballast base actually started smoking. The FEIT's had a pretty wide range of color temp, for being the same model.

    For organizations such as ourselves where we have areas that need to be lit 24/7/365, the savings are very easily calculated. In the 24/7 sockets, with myself and a student worker volunteering our time to purchase and install the bulbs, the cost of the bulb payed for itself in electric savings (city industrial rate, $0.141/kwh) in less than 5 weeks, over the 65W incan floods they replaced. Crazy.

  • by Guppy ( 12314 ) on Tuesday October 27, 2009 @01:29PM (#29886037)

    I only use them on the outside garage fixtures that our neighborhood covenant requires that I leave on all night. (They're on a light-sensing switch.) Despite the promises, they manage to only last about a year or two.

    There's your problem, light sensing switches (and dimmers) will absolutely destroy most CFLs. I'm surprised they lasted over a year. Your typical light sensing switch isn't equivalent to a regular light switch that flips on and off based on the amount of light.

    There's a couple of problems with photosensor switches. First, around dusk and dawn it may flicker on and off, which shortens the life of CFLs (but not cold-cathode CFLs, which are ok with rapid cycling). Second, even when completely "off", many photosensor switches will leak a bit of current, which may mess with your CFL's electronics, anything less than full-on / full-off is bad. Third, some photosensors and dimmers may have built-in "bulb saver" features meant to extend the life of incandescents -- they may pass the current through a diode or negativetemperature coefficient (NTC) thermistor (which again will kill CFLs).

  • by bradley13 ( 1118935 ) on Tuesday October 27, 2009 @02:52PM (#29887209) Homepage

    ...you just have to buy quality stuff. About 10 years ago, we bought five standing lamps, each with 3x32watt dimmable bulbs. The electronics in the lamp are specifically designed to dim CFLs, and the CFLs are designed to be dimmed. The total price for each lamp (they are nice lamps) was several hundred dollars. However, in 10 years, we have replaced only one bulb. The warm-up time is negligible and the light quality is excellent.

    Hot-wire bulbs are a throw-away product. You just can't look at CFLs the same way: you are buying an electronic appliance that ought to last for years. Either spend for quality, or use some other kind of lighting.

    You get what you pay for.

  • by Obfuscant ( 592200 ) on Tuesday October 27, 2009 @03:03PM (#29887379)
    Have a look at that dimmer switch and the wiring. Had the same problem myself, the problem turned out to be a bad ground wire. Incandescents had no problem in that fixture.

    There's a reason incandescents didn't have a problem there: they operate using hot and neutral. They pay no attention to ground. Neither does the dimmer switch deal with ground. Ground is a safety issue for humans.

    And CFLs operate exactly the same way. There is no ground connection on a CFL, just hot and neutral. They can't break due to a "bad ground" because they never touch ground.

    It's like saying your car gets bad gas milage because the diesel fuel in the truck parked next to it was contaminated.

    CFL fail miserably when using X10 controllers. They seem to have some current pulse that occurs after turnoff that makes the X10 controller think you are trying to turn the light back on using the local switch. Press X10 off -- click -- light off -- click -- light on! Press off again -- click off -- click on! It's like a video game, how many times do you have to press "off" to get them to stay off, and how short can you get the 'on' times to be?

    That, and the extremely short lives they have compared to simple incandescents, make them a pain in the ass and poor replacements. I like the european guy who talks about us americans and our "extravagant lifestyles" because we use incandescents. Using a 50 cent light bulb for ten years compared to ten (mercury containing) CFLs in the same place is extravagant?

  • Comment removed (Score:3, Informative)

    by account_deleted ( 4530225 ) on Tuesday October 27, 2009 @03:22PM (#29887633)
    Comment removed based on user account deletion
  • by Lorens ( 597774 ) on Tuesday October 27, 2009 @04:13PM (#29888385) Journal

    Aside - I have two traditional hard drives in my PC. They've been spinning almost-nonstop since 2003. Any idea how much longer I have until they crash?

    Until you stop them. They continue spinning well past the point where the wear will stop them from spinning up after stopping.

    Corollary: always make sure you have up-to-date backups before shutting off a long-running machine.

  • by Anonymous Coward on Tuesday October 27, 2009 @04:34PM (#29888707)

    A current cool white Cree MC-E LED can maintain a luminous flux of up to 730 lumens, that's about the same as a 60 watt (non-halogen) incandescent lamp. Still it requires less than 15W at this current. High-powered LEDs indeed have many advantages over CFLs. Some more that you didn't mention:

    They are *better* dimmable than incandescent lamps (this requires smart electronics though, you can't just use a potentiometer) because they run more efficient when underpowered instead of less. Running an incandescant lamp at 20% of the rated brightness will still require 50% or so of the rated power. Running an LED at 20% of the rated brightness will often require significantly *less* than 20% of the rated power.
    No wear even when switched on and off extremely rapidly.
    Gradual degradation instead of spontaneous failure. It was nice to have a replacement bulb with you everywhere you took your maglite, but LEDs make this unnecessary.
    Compared to CFLs: Slower to change hue (this is important when used as screen backlights).
    Takes up much less volume (the Cree Xlamp chip in some of my flashlights is indeed *tiny* but able to pump out up to 180 lumens).

    However it should be pointed out that there are huge quality differences between LED models as well. There is simply no way a no-name LED with plastic lens made in China will be able to output a useful amount of light after 50,000 hours, all bold claims to the contrary notwithstanding. E.g. the chip for a white LED emits some UV radiation that in the long term causes the plastic to cloud (sunlight aggravates the problem, obviously). A higher quality and much more expensive LED with a glass lens will not suffer from this problem.

    Now what was the article all about? ;-)

  • Re:Don't Defrag (Score:1, Informative)

    by Anonymous Coward on Tuesday October 27, 2009 @04:37PM (#29888765)

    Actually, that's not entirely true - sequential read and write are still faster than random read and write on SSDs, most likely due to caching and block access. There's not nearly the perf hit as on physical drives, but it's still there.
    http://www.anandtech.com/storage/showdoc.aspx?i=3531&p=3

  • by Anonymous Coward on Tuesday October 27, 2009 @09:18PM (#29892109)
    How about an efficient bulb made from...glass and tungsten?

    GE was working [businesswire.com] on a more-efficient version of the incandescent bulb. It was to start out twice as efficient as current bulbs and eventually be four times as efficient; comparable to CFLs. I'm guessing they planned to replace the tungsten filament with a tungsten photonic lattice [google.com]. Alas, stupid laws that banned incandescents outright -- instead of banning inefficient bulbs -- caused them to drop the project [cleantechnica.com].

    IMHO, CFLs suck. I've had nearly a dozen (various brands and price ranges) fail in the last 18 years. One of them (a ring-shaped bulb made by Lights Of America) came close to bursting into flames. Because of that, I no longer trust them.

    The article refers to bulbs made by Feit. I bought an LED night light made by them which failed after 1 year.

Two can Live as Cheaply as One for Half as Long. -- Howard Kandel

Working...