Most reliable SSD

[Darxide]

Which SSD would you guys say is the most reliable? Not necessarily the fastest.
I'm putting together a Windows Home Server, and for reasons of power and noise, I'd like to use an SSD as the boot drive.
It'll have other mechanical drives for storage on a RAID card, but they will only spin up when required.
Need a smallish, maybe 60Gb or thereabouts SSD.

The only experience I have with SSD's is an OCZ Agility that crapped out shockingly fast, and my current Crucial M4's, which I have to say have so far been perfect.

[andressergio]

Intel the best for me since early X25MG2 all the way up

[bluestang]

I don't think you can get much better than the Crucial M4, especially with the price you can get themat now.

[Computurd]

that is a loaded question. my money goes to intel OR the M4 as well. both are awesome recommendations. i would lean to M4 myself.

[iboomalot]

plextor is right up there and pretty fast too.

Not a fan of the 320 and the 520 is pretty expensive.

I own M4 , Plextor and Samsung

[Gilgamesh]

Perhaps Crucial is the better also for reability (or Crucial M4 win on Intel 320 and some 510 if you prefer :

http://translate.googleusercontent.c...asIZW8xHN9xI9A

"parts sold between April 1 and October 1, 2011, for returns created before April 2012, six months to one year of operation."

The statistics are based on mark by a minimum sample of 500 sales, with those models on a minimum sample of 100 sales, the largest volumes involving tens of thousands of parts by brand and thousands of pieces per model. Each time, we compared the rates obtained by the manufacturers to those of our previous article on the subject published in October 2011 .

- Crucial 0.82% (against 0.8%)
- Intel 1.73% (against 0.1%)
- Corsair 2.93% (against 2.9%)
- OCZ 7.03% (against 4.2%)

Crucial to take the top spot at the Intel for a significant increase in the rate of return for failure in the latter. It should be noted that the sample for this time Intel is just above the minimum required, and some of the return is related to " Bug 8 MB ​​"since resolved. OCZ return rate for failure explodes to 7%, and the manufacturer has exclusive references that exceed 5% failure rate:

- 15.58% 2 OCZ Vertex Series 240GB SSD
- 13.28% 2 OCZ Vertex Series 160GB SSD
- 11.76% 2 OCZ Vertex Series SSD 80GB
- 9.52% 2 OCZ Vertex Series 120GB SSD
- 8.57% 3 OCZ Vertex Series 120GB
- 7.49% 2 OCZ Vertex Series SSD 60GB
- 6.61% 2 OCZ Vertex Series 3.5 "SSD 120GB
- 6.37% 3 OCZ Vertex Series 240GB
- 6.37% March 60 GB OCZ Agility
- 5.89% 2 OCZ Vertex Series 100GB SSD

The Vertex monopolize the top two places but the Vertex 3 are not far behind. Note that the next time they behave much better because so far only reached 1.01% example of the Vertex 3120 Go, changes in the firmware is not foreign to this change.


Nothing can beat Crucial M4 128/256GB for price/performance/reability.

[Ao1]

I wonder why there are no stats for Samsung. Maybe they did not sell enough. Shame because I think the Samsung statistics would be right up there with Micron/ Intel.

From the conclusion:

“What about the future? We cannot predict, by cons, five products sold between October 1 2011 and 1 April 2012 for a return before April 2012 (0-6 months of use) that display the return rate the most important by category (minimum sample size of 100 pieces)”:

SSD:
13.46% Series 128 GB OCZ Petrol
5.95% 2 OCZ Vertex Series 3.5 "120 GB
5.85% Octane Series OCZ 128GB SATA II
4.57% 2 OCZ Vertex Series 120GB SSD
3.57% Octane Series OCZ 64GB SATA II

I’m surprised the Octane is not a lot higher as there are widespread reports of premature relocated sectors/ data loss. OCZ manages to single handedly give the SSD industry a bad name.

[Darxide]

Wow! Look at those figures for OCZ. That's abysmal.
I work in a quality control role myself, and I can tell you that with a failure rate like that, I'm amazed they are still in buisiness.

Looks like I'm buying another M4 then.

[RADCOM]

Ebuyer recently had an offer on the M4 so I bought another one, crucial r crucial:-)

[thegoatman]

I seem to recall from the endurance thread that the 256 GB Samsungs tend to wear out faster than the M4 or the Intel SSDs. Maybe I'm just reading the graphs wrong, though. Anyway, the Samsung is significantly faster, especially on uncompressable files.

[Christopher]

I seem to recall from the endurance thread that the 256 GB Samsungs tend to wear out faster than the M4 or the Intel SSDs. Maybe I'm just reading the graphs wrong, though. Anyway, the Samsung is significantly faster, especially on uncompressable files.

ummm... Not exactly. The 830 writes faster, so if you write as fast as you can for as long as you can, it will wear out faster. In the endurance test we are writing as fast as possible. The 830 is faster so it will write more per day. In actual usage you'd not be writing 24,000+GiB a day. The 256 830 made it to over 800TB before depleting MWI.

It only wears out faster if you want it to.

[thegoatman]

I guess I misunderstood the graph. Then again, I am not even sure I know what a GiB is. I can guess, but I am not really familiar with that unit. So maybe I should shut up.

My understanding was that the Sammie wore out more with less data written than its competitors. Was that wrong?

[kiwi]

Intel SSD 710 as it is targeted towards enterprise solutions and it is a successor of x25-E. We have a few x25-E running in the server environment for 3 years with 0 issues.

However, for a home server you would probably be fine with other brands as well (crucial or corsair are worth considering) or cheaper Intel 5xx or 3xx.

[minpayne]

I had a previous thread locked here. An updated version is posted there. My statistics is by far with the largest sample size specified explicitly. Use your own justification if my threads are used for reference.

My opinions: Intel 320 Series is still prone to be bricked into 8MB even with the updated firmware, so avoid it if you are not comfortable with sudden data loss. The M4 256GB and M4 512GB models are also risky. The Intel 520 Series is still prone to BSOD problems. The Plextor M3 Series is known to have some compatibility issues with Mac OS but doing fine with Windows. The Samsung 830 gets my vote.

[Christopher]

I guess I misunderstood the graph. Then again, I am not even sure I know what a GiB is. I can guess, but I am not really familiar with that unit. So maybe I should shut up.

My understanding was that the Sammie wore out more with less data written than its competitors. Was that wrong?

The 256GB 830 is still going. It isn't "worn out" but the "Life Left" percentage went from 100 to 1 in about 800TB. But the drive is still going well past that. Because it's a 256GB drive, it has to write a lot more to wear out, 4x times more than a comparable 64GB drive. It just writes really, really fast so it can write 2x the speed of the next fastest drive.

[Cyber-Mav]

avoid anything that uses sandforce or is made by OCZ and you should be good to go.

dang how bad could OCZ's rep get? everywhere i read it always about some ocz ssd dying out.

[[XC] Oj101]

I still don't agree with you on the Intel 320, I sold a total of 143 of them and two came back, neither for the 8 MB issue. One completely FUBAR (the PC hangs as soon as it is connected / doesn't POST if you boot with it connected), the other with a damaged power connector. I find it hard to believe that so many people are having an 8 MB issue when there aren't literally hundreds of thousands of reports online and I haven't come across it as of right now.

[Darxide]

avoid anything that uses sandforce or is made by OCZ and you should be good to go.

dang how bad could OCZ's rep get? everywhere i read it always about some ocz ssd dying out.

Well if I avoid Sandforce, that rules out a lot! But yeh I agree about OCZ. They have lost me as a customer for life! There's NO excuse for a drive that craps out after 20 minutes of use. Yes! That's how long my agility lasted. Didn't even get Windows fully installed when it BSOD'd becuase the HDD was 'missing'.

Anyway heres a quick question, has anybody any thoughts on the new Sandisk SSD's? They are slow yes, as they use the older SF-1222, but as I said, that doesn't matter to me.
And I can get a 120Gb for £3 more than a 64Gb M4. Really tempted.
Also, they guarantee it can withstand 80Tb's worth of data written to it. As a server OS disk, that will rarely be changed, that should last years!

[minpayne]

I still don't agree with you on the Intel 320, I sold a total of 143 of them and two came back, neither for the 8 MB issue. One completely FUBAR (the PC hangs as soon as it is connected / doesn't POST if you boot with it connected), the other with a damaged power connector. I find it hard to believe that so many people are having an 8 MB issue when there aren't literally hundreds of thousands of reports online and I haven't come across it as of right now.

Then why is this thread being bumped again and again? Of course it is possible that other SSD brands/companies have employed online actors to demote Intel's product but it is hard to believe that the feedbacks from verified owners at newegg.com are giving false information. While your sample size is smaller than the newegg sample size, I could only say that the newegg stats gives more power.

My own Intel 320 Series 600 GB is still living, but I do daily backup of important data.

[Christopher]

Well if I avoid Sandforce, that rules out a lot! But yeh I agree about OCZ. They have lost me as a customer for life! There's NO excuse for a drive that craps out after 20 minutes of use. Yes! That's how long my agility lasted. Didn't even get Windows fully installed when it BSOD'd becuase the HDD was 'missing'.

Anyway heres a quick question, has anybody any thoughts on the new Sandisk SSD's? They are slow yes, as they use the older SF-1222, but as I said, that doesn't matter to me.
And I can get a 120Gb for £3 more than a 64Gb M4. Really tempted.
Also, they guarantee it can withstand 80Tb's worth of data written to it. As a server OS disk, that will rarely be changed, that should last years!

SanDisk's new Extreme SF2281s with Toggle NAND are not slow at all.

[Darxide]

SanDisk's new Extreme SF2281s with Toggle NAND are not slow at all.

Yeh sorry, I meant the 'ultra', I didn't know they'd been out for a while, they only just appeared on the store I use.
I bought it anyway, it actually just arrived like an hour ago but I have to work tonight so no time to install it :-(
I got the 120Gb version for £70. It uses last gen SF1222, and is only good for 280 read and 270 write, but like I said, it's for a server OS disk, so speed is not what I'm after.

[[XC] Oj101]

Then why is this thread being bumped again and again? Of course it is possible that other SSD brands/companies have employed online actors to demote Intel's product but it is hard to believe that the feedbacks from verified owners at newegg.com are giving false information. While your sample size is smaller than the newegg sample size, I could only say that the newegg stats gives more power.

My own Intel 320 Series 600 GB is still living, but I do daily backup of important data.

That thread has 68 posts, hardly a startling figure. How many 320 Series SSDs were sold? Lets work on only 5,000 being sold - that would still only be a 1.36% failure rate - not the 50 % or more that it's made out to be.

[johnw]

Here are the Intel SSD return rates for a large French retailer, for SSDs which were sold during the 6-month period starting in the date shown, and returned within 1 year of the starting date shown:

0.59% 2009-Oct-01
0.3% 2010-Apr-01
0.1% 2010-Oct-01
1.73% 2011-Apr-01

Note how the return rate increased sharply for Intel SSDs that started selling in April 2011. That corresponds almost exactly with when the Intel 320 started selling, replacing the G2 X25-M which had such low return rates. Now, 1.73% isn't bad, but it is significantly worse than the 0.3% for the corresponding period 1 year earlier when the G2 X25-M was the largest volume model for Intel. It would seem that the Intel 320 is about 6 times more likely to be returned than a G2 X25-M.

I think it is obvious, from the thread minpayne referenced, that there are still some bugs that cause the Intel 320 to panic and go into 8MB mode. It seems that many issues can cause such a panic, so there is not just one "8MB bug", but rather a whole class of them. Intel has obviously not fixed them all, but the main question is how prevalent they are.

Unfortunately, we will have to wait until October to see the statistics for return rate after Intel released the firmware update (Aug 2011) that supposedly fixed the "8MB bug".

http://www.behardware.com/articles/8...rns-rates.html
http://www.behardware.com/articles/8...rns-rates.html
http://www.behardware.com/articles/8...s-rates-5.html
http://www.hardware.fr/articles/862-7/ssd.html

[Ao1]

There is some interesting research here: (nothing to do with RMA's)

http://www.eetimes.com/design/memory...ash-based-SSDs

I’m even more wary of SSD’s with large caches now. My 830 has been fine and I even experimented cutting the power and it did not cause problems, but as the cache seems to be dealing with mapping out logical block addresses with physical flash memory locations it would be theoretically possible to lose all your data in the blink of an eye. If anything goes wrong with the cache it doesn't matter how good the intergrity of the data is on the NAND.

I don’t know if a BSOD counts as a power failure, but if it does it makes me wonder if f/w bugs that cause BSOD are causing collateral damage in the process and making things worse.

One really interesting thing from that research that I did not know: “Each MLC cell contains two bits of data, and each bit belongs to a different logical NAND page”

[Cyber-Mav]

anyone got controller based stats to see which is more failure prone?

would be good to see whats failing more out of sandforce, marvell, indilinx, jmicron, intel, etc

im guessing sandforce will top the rest by several orders of magnitude.