I am lazy so...SSD users, please reply

[Snowhog]

They were out of KIngston Hyper X drives so I settled for a pair of Intels, which coincidentally have the Sandforce controllers. Trust me, if they 'kahck' before their touted 5 year warranty, I will be showing up at their door, lol.

http://ark.intel.com/products/66248

Intel® SSD 520 Series
(120GB, 2.5in SATA 6Gb/s, 25nm, MLC)

Mean Time Between Failures (MTBF)

1,200,000 Hours

That's 136.9 years! Does anyone really believe that?!!

[tek_heretik]

Intel® SSD 520 Series
(120GB, 2.5in SATA 6Gb/s, 25nm, MLC)

Mean Time Between Failures (MTBF)

1,200,000 Hours

That's 136.9 years! Does anyone really believe that?!!

Nope.

[SteveRiley]

MTBF is a statistical measurement. By only reporting hours, Intel is intentionally trying to confuse people.

From https://www.ece.cmu.edu/~ganger/ece5...f.description:

MTBF is the inverse of the failure rate in the constant failure rate phase. Nothing more and nothing less. The units of MTBF are (or, should be) units of "thing-time" pre failure; e.g. machine-hours per failure or system-years per failure but the "thing" part and the "per failure" part are almost always omitted to enhance the mystique and confusion and to make MTBF appear to have the units of "time" which it doesn't. We will bow to the convention of speaking of MTBF in hours or years -- but we all know what we really mean.

What does MTBF have to do with lifetime? Nothing at all! It is not at all unusual for things to have MTBF's which significantly exceed their lifetime as defined by wearout -- in fact, you know many such things. A "thirty-something" American (well within his constant failure rate phase) has a failure (death) rate of about 1.1 deaths per 1000 person-years and, therefore, has an MTBF of 900 years (of course its really 900 person-years per death). Even the best ones, however, wear out long before that.

This example points out one other important characteristic of MTBF -- it is an ensemble characteristic which applies to populations (i.e. "lots") of things; not a sample characteristic which applies to one specific thing.

[Snowhog]

A "thirty-something" American (well within his constant failure rate phase) has a failure (death) rate of about 1.1 deaths per 1000 person-years and, therefore, has an MTBF of 900 years

So finally, the Biblical age of Methuselah is explained!!

[SteveRiley]

LOL, nice one. Now, go show up at your local church and make the case that the bible is not reporting actual age, but instead reporting human MTBF. I would pay to watch that spectacle

[oshunluvr]

For me, the only actual number that means anything real is the warranty because that is the indicator that determines when I have to replace something at my own expense. Once SSD warranties matched or exceeded those of hard drives, I was in the market for one (I own three now). Yes, yes, yes - the inconvienence of mailing it back in blah blah blah. Still an unlikely event and one I'm willing to risk.

What I find really odd is the return rate (percentage of drives returned) for SSD's is less than half of the rate for hard drives (looking at the 4 lowest manufacturers return rates in both categories). This means you are at least twice as likely to have a drive failure during the warranty period with a hard drive. This is both logical (no moving parts - no breakage) and information easily found on the web in a few milliseconds. And yet: I still read people posting statements like "Well, if you want reliability, get a hard drive." A totally unsupportable statement.

My last hard drive failures occured about 4 years ago. I had 2 320GB WD drives in a RAID. They failed within a few weeks of each other and (of course) a couple of months outside the warranty period. The replacements, now also outside the warranty period (4 500GB WD "Blue" drives) are still in service. I can honestly say I have discarded or donated more hard drives due to desire to upgrade than actual failure, so even at the doubled rate of failure - it ain't that bad in the real world.

IMO, the only reason to use a platter drive of any type is MB per dollar. I.e. my server = 6TB - NOT SSDs. My laptops/desktop = 32GB to 512GB - SSDs.

[dmeyer]

While I appreciate your real-world experience with SSD longevity, could you please explain the logic underlying the above conclusion? By any measure I can come up with (or anyone else on the 'net I've seen so far) Intel drives are middle-of-the-road. Again, considering real-world uses: Which areas does Intel excel in? Speed? Cost per megabyte? Reliability?

I've yet to read any review or comparison where Intel came out on top in any category. The 4 year-old X-25 is slower than a platter drive at Sequential Write Performance, which is probably why they switched to Sandforce controllers. Both Samsung and Plextor have lower return rates than Intel on the order of a decimal place. Most of the "short lists" for performance or cost don't include any Intel models because they're too slow and too expensive.

I submit the above opinion is in the same league as the SSD life expectancy opinion: Based wholly on out-dated information and Internet FUD.

Hmm, ok, maybe that viewpoint may not be true anymore but when SSDs first came out and when they did fail miserably all the time, I remember Intel SSDs being incredibly reliable when others where ticking time bombs. They have never been very fast, granted, but they were bulletproof. It's very possible that this has changed in the past few years but that initial impression of reliability has always stuck with me and to me a reliable HDD/SSD is the best HDD/SSD because of the value of my data.

On top of that, my anecdotal experience on my SSDs. those of family and friends, being on a number of forums and just general university life in class and out reflects this initial impression I have seen a decent number of SSDs die and none of them has ever been an Intel drive (none was a result of flash wear as far as I know so my view on longevity in terms of wear stands). Now this is anecdotal so how true a reflection it is, is very difficult to say.

So yes, my opinion is that Intel SSDs are the best but the basis for my evaluation is probablly different to most. They aren't the best value for money, nor are they fastest but they have been, anecdotally the most reliable and that is what I place the greatest value.

On top of all this, I would also like to say, consumer data is pretty terrible to measure the reliability of higher end parts like SSDs. A good example is a university that orders a large number of SSDs directly from a manufacturer (and only recently has the majority of SSDs been bought by individuals and not institutions), now when they hit an issue they negotiate directly with the manufacturers. Due to all this skewed reporting is inevitable especially considering Intel is usually not a consumer's first choice due to priceerformance ration then it can skew even worse. Also are those return rates based on device failure or does it include buyer regret?

So in conclusion, my viewpoint might be wrong but I stand by it because there has never been a proper study into the flash memory industry (a well known and oft criticized fact) thus my anecdotal experience is the only one I can trust .

[oshunluvr]

Again, I appreciate your actual experience over those of techie websites and industry hype machines. On the other hand - in real world use (that's my world of course), the outcome of an industrial or institutional use of an SSD has rather little meaning. If I were producing millions of dollars of income with my machine and moving data all day every day, I would obviously give up a few minutes or even hours a day of speed for quantifiable reliability, but this is of course not the case. The dollars I'm spending are for performance and nothing else. The tolerance threshold for failure is much higher here.

I hadn't considered the return data including remorse returns, but even so, SSDs fair better by simple math. The stat's I reviewed are published by a large French online retailer and incompasses their returns delclared as defective by their tech department.

[SteveRiley]

My main laptop, the T520, has a Crucial M4 256 GB SSD. It's nearly a year old. Performance numbers today are the same as when it was new:

steve@t520:~$ [B]sudo hdparm -tT /dev/sda[/B]

/dev/sda:
 Timing cached reads:   13546 MB in  1.99 seconds = 6792.66 MB/sec
 Timing buffered disk reads: 1144 MB in  3.00 seconds = 381.31 MB/sec

My X1 came with a Lenovo-branded Toshiba SSD that was dog-slow. I replaced it with a 60 GB Intel X-25 I had, which performs much better.

I've been very happy with the Crucial, and would likely use my own money to buy their stuff. I know my IT department is pleased. BTW, costs are really coming down -- Crucial now sells a 960 GB SSD for $599! That's exactly what I paid for an 80 megabyte full-height hard drive in 1987. Ugh.

[tek_heretik]

Don't take this the wrong way, if you guys want to keep hashing this out, by all means, lol, but the idea was to give an aging system some more speed, pep and longevity, not only that, it's using less electricity, creating less heat (reducing the 'thermal pollution' inside the case), making less noise, waaaaaaaaaaaaaay less noise (I was never fond of the subtle buzzy whine of HDDs). By the time this machine is ready for the proverbial 'heap', the i9 or i11 will be out, then I will build a whole new box.

I think we have reached the point in history (most likely breifly) where the exponential bloat factor of demanding software has slowed to a crawl and isn't overtaking hardware capabilities by leaps and bounds anymore.

Edit: But for backup storage, I'll stick with magnetic platters for now...thank you very much.

[tek_heretik]

I did an experimental shred of an old TV show video file, gkrellm says the overwrites (35 to be exact) were averaging 385MB/s, not too shabby, that's a 100 or so more than before with the 4 HDD RAID 0.

[dmeyer]

Again, I appreciate your actual experience over those of techie websites and industry hype machines. On the other hand - in real world use (that's my world of course), the outcome of an industrial or institutional use of an SSD has rather little meaning. If I were producing millions of dollars of income with my machine and moving data all day every day, I would obviously give up a few minutes or even hours a day of speed for quantifiable reliability, but this is of course not the case. The dollars I'm spending are for performance and nothing else. The tolerance threshold for failure is much higher here.

I hadn't considered the return data including remorse returns, but even so, SSDs fair better by simple math. The stat's I reviewed are published by a large French online retailer and incompasses their returns delclared as defective by their tech department.

Yeah, I also valued performance highly and bought a Corsair Force GT. Really good value for money and so fast. Hope I didn't make you think that you should go for Intel, I was just giving my anecdotal experience combined with the typical "reliability" first mindset of many Linux users. Plus, as a student, I can't really afford the Intel stuff. I can highly recommend Crucial and Corsair SSDs. I've had only good experiences with them.

[Guest]

To affirm this viewpoint and somewhat disarm those who believe wear is a serious issue on SSDs I'll provide some fairly anecdotal evidence. Friend had a used Corsair Force GT (I have the same one) and after a year he replaced it with newer Intel ones (Intel make the best SSDs on the market by a large margin) and wanted to clear it before selling it. Now the Corsair Force GT runs on Sandforce controllers which are notoriously bad on the drives. He was a little paranoid about wiping the drive and because securely deleting files off a SSD and flash memory generally isn't well studied we cam eup with a simple solution: write 1's the 0's and repeat a few times. Boot a live disc, set up some scripts and we wrote 1's followed by 0's for a few days straight on his 180GB drive. Considering these were raw operations, we were getting 450+ MB/s comfortably. Realistically we must have written at least a couple hundred TBs to that drive.

Now if we were talking server side stuff, using anything other than SLC would be dumb.

I am a happy user of Corsair Force GT which comes with the sandforce controller. I am not sure why someone needs to be worried for wiping off the drive contents on a sandforce controller. I have performed this several times in my test lab & trust me never had any bad experiences. Unless you do not have the latest firmwares, sandforce SSDs never put into trouble. This is my personal exp & always happy to share this with my tech friends.

[dmeyer]

I am a happy user of Corsair Force GT which comes with the sandforce controller. I am not sure why someone needs to be worried for wiping off the drive contents on a sandforce controller. I have performed this several times in my test lab & trust me never had any bad experiences. Unless you do not have the latest firmwares, sandforce SSDs never put into trouble. This is my personal exp & always happy to share this with my tech friends.

It's simple. People aren't exactly sure how to properly securely delete off an SSD. With a traditional spindal drive we know that 7 overwrites tends to be sufficient that the data is absolutely irrecoverable and there are specific ways to delete which extends well past just removing the data pointers. Researchers are still currently studying the best way to delete data off SSDs because of how fundamentally different they are. So put simply, deleting is often not enough to get rid of the data, often not even overwriting is sufficient, you need to securely delete your data and that's not well studied on an SSD therefore we sort of just tried to fill it with so much that we felt comfortable that nothing was recoverable.

[tek_heretik]

I didn't purposely go Intel, they were out of the 'on sale' Kingston Hyper X drives, which incidentally have 2 years LESS warranty, it's all good, mission accomplished, that's all that matters.