[TR] SSD endurance experiment ends at 2.4 PB

[Essence_of_War]

http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead

The 840 pro is the last man standing. Excellent showings from all participants, though!

 

[Emulex]

Yeah! I have my 830 and 840 pro's rocking almost 2 years under ESXi through megaraid controllers running hardcore SQL server databases! Solid!

 

[Insert_Nickname]

The 840 pro is the last man standing. Excellent showings from all participants, though!

Having followed this experiment, I'm really impressed by all the drives. The 840 pro lasting 2.4PB of writes is just mind-boggling...

 

[Elixer]

Really too bad that they didn't have a bigger sample size.

 

[Hulk]

I don't think that write endurance is going to be a problem for most users. It's data retention which might be the elephant in the room.

And that ain't so easy to test. Take 5 drives.
Don't use one.
Wear the 2nd to 25% media wear indicator.
The 3rd to 50% media wear indicator.
The 4th to 75%...
The 5th to 100%

As soon as the data has been written (or immediately for the 1st drive) run HD Tach, then run every week there after for 5 years.

Then we'll know more about data retention and endurance!!!

Yeah I know that's crazy.

 

[Charlie98]

I can attest to the 840Pro failing suddenly and without warning...

 

[Cerb]

I don't think that write endurance is going to be a problem for most users. It's data retention which might be the elephant in the room.

Yes, but now we can point the people that still worry about it to this. Plus, this experiment, by comments, appears to have turned many opinions around.

 

[razel]

Samsung knows how to bin their flash. They have one of the earlier drives to fail and one of the last. Regardless, it's time to bust the myth of SSD write endurance and for everyone to wisely ignore web guides and articles that continue to preach various 'tweaks' to lessen writes with SSDs.

When buying one, just use it and if it bothers you just double check that performance is up to snuff.

 

[Emulex]

The bad part of SSD's are like the Intel 335 which self suicide at the wear mark, or those drives SANDFORCE with lifetime throttling! Samsung 830/840 pro and hopefully the 850 pro are just a testament to reliability!

 

[Hulk]

The bad part of SSD's are like the Intel 335 which self suicide at the wear mark, or those drives SANDFORCE with lifetime throttling! Samsung 830/840 pro and hopefully the 850 pro are just a testament to reliability!

Please explain?

 

[Emulex]

Read the article on how the intel 335 died! "Intel's 335 Series failed much earlier, though to be fair, it pulled the trigger itself. The drive's media wear indicator ran out shortly after 700TB, signaling that the NAND's write tolerance had been exceeded. Intel doesn't have confidence in the drive at that point, so the 335 Series is designed to shift into read-only mode and then to brick itself when the power is cycled. "

And some sandforce drives had lifetime throttling which reduced throughput according to the warranty (from previous article on SSD endurance test) So you would write to the SSD for 3 months straight and it would then punish you for the next 21 months at 1K/sec write speeds to protect its wear!

 

[Riceninja]

this really shuts the whole MLC vs TLC argument. it takes more than 400TB for the ssds to deteriorate and my 256gb 840 basic has barely 7TB written after 1.5 years. and it doesn't even need to last much longer - ive already got a 500gb 850 evo on the way (cheaper than what i paid for the 840 as well).

technology is moving far too fast for me to care about long term endurance.

 

[Hulk]

Read the article on how the intel 335 died! "Intel's 335 Series failed much earlier, though to be fair, it pulled the trigger itself. The drive's media wear indicator ran out shortly after 700TB, signaling that the NAND's write tolerance had been exceeded. Intel doesn't have confidence in the drive at that point, so the 335 Series is designed to shift into read-only mode and then to brick itself when the power is cycled. "

And some sandforce drives had lifetime throttling which reduced throughput according to the warranty (from previous article on SSD endurance test) So you would write to the SSD for 3 months straight and it would then punish you for the next 21 months at 1K/sec write speeds to protect its wear!

I did read the article. I have a different interpretation. Intel has very high reliability standards. 700TB is like 10 times the rated endurance. They are trying to protect the user from a catastrophic failure.

 

[Puffnstuff]

I can attest to the 840Pro failing suddenly and without warning...

Unfortunately I can also say that the 840 pro failed without warning and I was not a happy camper when it happened. My usage was not anywhere near those on the test when it happened. I can only hope that improvements to the drives will continue to enhance their performance and longevity.

 

[Charlie98]

technology is moving far too fast for me to care about long term endurance.

True that. I think we are at the point of just buying an SSD and using it... with the likelihood that it will be replaced by a bigger (or faster, if you are a speed demon, or if 'SATA4' emerges) drive loooong before the wear limit is reached. Once I got away from my original 64GB SandForce OCZ SSD, I pretty much quit spoon-feeding my SSD and just went back to work; I load everything on it and run with it.

 

[greenhawk]

Read the article on how the intel 335 died! "

And some sandforce drives had lifetime throttling which reduced throughput according to the warranty (from previous article on SSD endurance test) So you would write to the SSD for 3 months straight and it would then punish you for the next 21 months at 1K/sec write speeds to protect its wear!

it sucks the intel did that. I under stand why it would do it, but as a consumer I would want the ability to keep going (ie: a pause in the boot once reached it's limit on every power up). But going dead without warning is not acceptable to me for a drive sold to the general market. If being sold as a server grade device, fair enough.

Re sandforce, that is just dirty. Care more about making it to the warranty expiry more than given the consumer what they paid for / expect. I have not been a fan of sandforce controllers (most of my usage patterns are not friendly to compression), but that warranty idea just makes me think even less of them. Might was well add them to the group of companies that put page counters in the ink cartridges of printers to "kill" them after a set number of pages are printed, even if there is still ink left, or even by selling printers with "starter" cartridges which make the whole new printer cheaper than a single replacement cartridge.


/end rant

 

[SSBrain]

Putting the data retention issue aside (although at the rated P/E limit which all SSD can reach it's supposed to be at least 1 year of time for consumer drives) I can't really take write endurance tests seriously when the write amplification isn't also taken into account. It can be misleading to say that a drive lasts 'xxx' terabytes written (or also "only" when results appear low). It depends on the typical workload/usage (or in other words, how they set up the tests), and thus write amplification.

In the end they're gauging NAND quality, so it's the amount of P/E cycles performed on average that they should be interested in.

 

[mikeymikec]

I can attest to the 840Pro failing suddenly and without warning...

Just because TR ran endurance trials and certain drives lasted for x amount of writes, it doesn't not mean that yours will last that long and die in the same way.

Hands up who has had a HDD fail without warning, then hands up who's had a HDD fail in the slow and steady way.

Furthermore, I wouldn't label a drive as "dying suddenly" if its drive health had said zero percent for absolutely yonks. How much warning do you want exactly?

 

[beginner99]

IMHO the Intel drive is the winner. It warned you. It gave you the chance to create a fully functional image/backup and then stopped working hindering you to create stuff you could never back-up.

 

[Emulex]

Most people don't notice SMART warnings until they reboot their pc. thus rendering the intel warning system moot.

I've had an intel x25-m do just this, go into read only mode, only to crash the system, come back, on-boot warn me of a smart imminent failure, and be kaput 100% trashed!

 

[mikeymikec]

Most people don't notice SMART warnings until they reboot their pc.

That depends, I believe. I'm not sure exactly what this depends on, but I have seen a Windows Vista laptop throw an error during a Windows session that the disk is likely to fail (I think because it reported its SMART status as BAD possibly, but my experience of 'BAD' is usually 'failed utterly', with a few exceptions, having said that I don't often see 'BAD').

 

[Insert_Nickname]

Hands up who has had a HDD fail without warning, then hands up who's had a HDD fail in the slow and steady way.

You can sign me up for both. Though catastrophic failure is more common in my experience.

Really the only certainty is that a drive will eventually fail. Its just a matter of time or bad luck. That goes for both HDDs and SSDs.

Therefore the mantra: backup, backup and backup...

 

[Charlie98]

Just because TR ran endurance trials and certain drives lasted for x amount of writes, it doesn't not mean that yours will last that long and die in the same way.

Hands up who has had a HDD fail without warning, then hands up who's had a HDD fail in the slow and steady way.

Furthermore, I wouldn't label a drive as "dying suddenly" if its drive health had said zero percent for absolutely yonks. How much warning do you want exactly?

Don't get me wrong, I'm not full of Samsung hate... I'm just conveying my experience. I had no warning.... was working in QuickBooks and the system locked up. No amount of reboot/different port/different computer would allow me to access the drive. It was GONE. Soooo... just a teensy bit of warning would have been nice... This was after I did a SMART check and performance optimization (with Magician) that very morning that gave me no indication of pending failure.

I found the OP article pretty interesting... I'm guessing my drive experienced a big hit of flash failures or uncorrectable data or something and just shut down. It was less than a year old and probably had less than 1TB of writes to it.

 

[IronWing]

Why do the drives brick up when no longer writable? In particular, why does Intel instruct the 335 drive to brick? I understand sending the drive into read only mode but to force the drive to fail just seems an unnecessary action that is not consumer friendly.

 

[johny12]

And some sandforce drives had lifetime throttling which reduced throughput according to the warranty (from previous article on SSD endurance test) So you would write to the SSD for 3 months straight and it would then punish you for the next 21 months at 1K/sec write speeds to protect its wear![/QUOTE]


That warranty period comment got me worried, but after reading around on some other sandforce controller SSDs it seems that setting is not universal and most of the SSDs with that controller don't do that.