Tech Report's SSD Endurance Test Is Down to Two

Subject: General Tech, Storage | September 21, 2014 - 05:41 PM |
Tagged: ssd, Samsung, kingston hyper x, kingston, endurance, corsair neutron gtx, corsair, 840 pro

Many drives have died over the last year and a bit. The Tech Report has been torturing SSDs with writes until they drop. Before a full petabyte of data was written, three of the six drives kicked the bucket. They are now at 1500TB of total writes and one of the three survivors, the 240GB Corsair Neutron GTX, dropped out. This was a bit surprising as it was reporting fairly high health when it entered "the petabyte club" aside from a dip in read speeds.

The two remaining drives are the Samsung 840 Pro (256GB) and Kingston HyperX 3K (240GB).

techreport-deadssds.jpg

Two stand, one fell (Image Credit: Tech Report)

Between those two, the Samsung 840 Pro is given the nod as the Kingston drive lived through uncorrectable errors; meanwhile, the Samsung has yet to report any true errors (only reallocations). Since the test considers a failure to be a whole drive failure, though, the lashings will persist until the final drive gives out (or until Scott Wasson gives up in a glorious sledgehammer apocalypse -- could you imagine if one of them lasted a decade? :3).

Of course, with just one unit from each model, it is difficult to faithfully compare brands with this marathon. While each lasted a ridiculously long time, the worst of the bunch putting up with a whole 2800 full-drive writes, it would not be fair to determine an average lifespan for a given model with one data point each. It is good to suggest that your SSD probably did not die from a defrag run -- but it is still a complete waste of your time and you should never do it.

Source: Tech Report

Tech Report's SSD Endurance: The Petabyte Club

Subject: General Tech, Storage | June 16, 2014 - 10:38 PM |
Tagged: ssd, Samsung 840, Samsung, kingston hyper x, kingston, endurance, corsair neutron gtx, corsair

In The Tech Report's ongoing SSD endurance challenge, three SSDs are soldiering forward. We have reached the thousand-terabyte mark, which is at least five times more than any of the survivors are rated for. These survivors: The Corsair Neutron GTX, the Samsung 840 Pro, and the Kingston HyperX 3K. Technically, the HyperX was able to reach 1PB of written data with performing only 716TB of actual writes, due to compression.

techreport-endurance-1000tb-speed.gif

Image Credit: The Tech Report

Of course, each of the drives are less-than prestine. The Corsair Neutron GTX 240GB was slowly decreasing in its "life" attribute since the beginning, claiming to be somewhere between 75% and 80% with a fairly linear decline. If this trend continues, the drive will reach "zero" at around 4-5PB of writes. That said, its read speed has substantially dropped from the time between 900TB and 1000TB of total writes, from 500MB/s to just under 400MB/s. Also, this "life" could drop substantially if the drive encounters reallocated sectors (which this model has apparently yet to do).

The other two drives are a similar, remarkably successful story.

The Kingston HyperX drive is reporting itself to be substantially worse off, within the last 10% of its life. That said, even though it claims to be pining for the fjords, it is still working and has only reported a couple of reallocated sectors, those occurring in the last 100TB of writes.

The Samsung 840 Pro seems to still be going strong, although it had more zero or "a couple" of reallocated sectors -- every hundred terabytes yields about 500 reallocations.

As always, this is just our brief discussion of what The Tech Report found out. Be sure to check out their full article for many more benchmarks, tests, and conclusions.

Source: Tech Report

Heavy peta-ing; fondling SSDs in a bad way

Subject: Storage | January 10, 2014 - 12:37 PM |
Tagged: ssd, petabyte, SSD test, endurance

The Tech Report's attempts to test SSDs to destruction have hit the 500TB mark, with three two-bit MLC NAND drives and one three-bit TLC model all trying to survive.  They are using raw SMART data to keep track of sectors reallocated from the spare area to replace flash which has died due to repeated usage.  So far the Samsung 840 with its three bit TLC has suffered the most loss of sectors but like the other drives it has not shown much performance degradation.  There have been a few other bumps in the road during the tests, check out the full story here.

angle.jpg

"Our SSD Endurance Experiment has reached the half-petabyte mark, so it's time for another checkup."

Here are some more Storage reviews from around the web:

Storage

(TechReport) SSD Endurance Gets Interesting at 300TB

Subject: General Tech, Storage | November 26, 2013 - 12:46 AM |
Tagged: ssd, endurance

Update 11/26/2013 @ 11:20pm Allyn weighed in, particularly about the Samsung drives. One thing that he notes (as does Tech Report in the original article) is that Samsung might just be more severe in its reporting of errors. For instance, he believes that it is just about impossible for SSDs to write 300TB of data without ever seeing a flash read error. Regardless of what they report, each of these drives keep on ticking. It is "significantly beyond the expectations of a consumer (non-Prosumer) SSD".

The Tech Report has been testing a batch of SSDs for their life expectancy over several months now. Results have been fairly interesting: drives were relatively stable even up to 200TB of cumulative writes; some drives even got faster. Now they have passed the 300TB threshold and we are seeing certain drives hit some harsh realities. As Scott Wasson said in a tweet:

 

 

In all, though, even the TLC offerings have surpassed reasonable expectations. Consumer drives are designed for consumer machines and will likely take decades to reach the hundreds-of-terabytes order of magnitude for today's usage.

techreportreallocated.gif

Image Credit: The Tech Report

It makes me wonder exactly how over-provisioned enterprise hardware is if these, consumer, parts have such high reliability.

Source: Tech Report

The continuing Tech Report SSD torture test

Subject: Storage | November 4, 2013 - 11:53 AM |
Tagged: ssd, endurance

The Tech Report have hit the 200TB mark on their testing of ~250GB SSDs from Corsair, Intel, Samsung and Kingston and the drives are starting to feel it.  At the 100TB mark Intel and Samsung drives started to lose blocks of storage and at 200TB all but two drives have shown evidence of degradation.  The non-Pro Samsung 840 has suffered the most but its performance is very similar to what it was in the beginning while the Corsair and the Kingston drive receiving only compressed data report themselves in perfect health.  Check out the exact performance deltas in their article.

TR_drives.jpg

"We're in the process of hammering six SSDs with an unrelenting torrent of writes to see what happens as the flash wears out. Today, we check in on the drives after 200TB of writes."

Here are some more Storage reviews from around the web:

Storage

22 Terabytes of hammering on the Anvil

Subject: Storage | September 6, 2013 - 11:25 AM |
Tagged: endurance, ssd, anvils storage utility

The Tech Report have seen some mixed results from their SSD endurance testing using Anvil's Storage Utility.  There has not been any mentionable performance degradation for any of the SSDs they have been testing but Kingston's drives have shown some unpredicted behaviour. The HyperX series displayed speed increases, a slight increase in sequential reads and writes as well as random writes and a large increase in random reads.   Tune in next time when they reach 100TB.

TR_anvil.jpg

"We're testing six SSDs to see how many writes they can take before burning out and what happens to performance as the flash degrades. Today, we check in on our subjects after 22TB of writes."

Here are some more Storage reviews from around the web:

Storage

A new way to kill your SSD

Subject: Storage | August 22, 2013 - 12:37 PM |
Tagged: ssd, endurance, anvils storage utility

The Tech Report is currently testing several SSDs to destruction, or at least trying to.  They are using a new tool called Anvil's Storage Utilities which includes a test designed to determine the longevity of the flash storage inside SSDs.  They started with factory fresh SSDs, never having a bit written to them before and are currently writing to every address on those drives with a goal of 22TB to be written before they test the speeds of the drives again.  Will some fare better than others?  Perhaps some will sacrifice capacity to keep their speed up?  Stay tuned, even with SATA 6Gbps it takes a while to write that much data!

software-anvil.jpg

"We all know that flash memory has a limited tolerance for write cycling, but what does that mean for SSD endurance? We're testing six SSDs to failure to find out."

Here are some more Storage reviews from around the web:

Storage

Subject: Editorial, Storage
Manufacturer: Various
Tagged: tlc, ssd, slc, mlc, endurance

Taking an Accurate Look at SSD Write Endurance

Last year, I posted a rebuttal to a paper describing the future of flash memory as ‘bleak’. The paper went through great (and convoluted) lengths to paint a tragic picture of flash memory endurance moving forward. Yesterday a newer paper hit Slashdotthis one doing just the opposite, and going as far as to assume production flash memory handling up to 1 Million erase cycles. You’d think that since I’m constantly pushing flash memory as a viable, reliable, and super-fast successor to Hard Disks (aka 'Spinning Rust'), that I’d just sit back on this one and let it fly. After all, it helps make my argument! Well, I can’t, because if there are errors published on a topic so important to me, it’s in the interest of journalistic integrity that I must now post an equal and opposite rebuttal to this one – even if it works against my case.

First I’m going to invite you to read through the paper in question. After doing so, I’m now going to pick it apart. Unfortunately I’m crunched for time today, so I’m going to reduce my dissertation into the form of some simple bulleted points:

  • Max data write speed did not take into account 8/10 encoding, meaning 6Gb/sec = 600MB/sec, not 750MB/sec.
  • The flash *page* size (8KB) and block sizes (2MB) chosen more closely resemble that of MLC parts (not SLC – see below for why this is important).
  • The paper makes no reference to Write Amplification.

Perhaps the most glaring and significant is that all of the formulas, while correct, fail to consider the most important factor when dealing with flash memory writes – Write Amplification.

Before geting into it, I'll reference the excellent graphic that Anand put in his SSD Relapse piece:

writeamplification2.png

SSD controllers combine smaller writes into larger ones in an attempt to speed up the effective write speed. This falls flat once all flash blocks have been written to at least once. From that point forward, the SSD must play musical chairs with the data on each and every small write. In a bad case, a single 4KB write turns into a 2MB write. For that example, Write Amplification would be a factor of 500, meaning the flash memory is cycled at 500x the rate calculated in the paper. Sure that’s an extreme example, but the point is that without referencing amplification at all, it is assumed to be a factor of 1, which would only be the case if you were only writing 2MB blocks of data to the SSD. This is almost never the case, regardless of Operating System.

After posters on Slashdot called out the author on his assumptions of rated P/E cycles, he went back and added two links to justify his figures. The problem is that the first links to a 2005 data sheet for 90nm SLC flash. Samsung’s 90nm flash was 1Gb per die (128MB). The packages were available with up to 4 dies each, and scaling up to a typical 16-chip SSD, that only gives you an 8GB SSD. Not very practical. That’s not to say 100k is an inaccurate figure for SLC endurance. It’s just a really bad reference to use is all. Here's a better one from the Flash Memory Summit a couple of years back:

flash-1.png

The second link was a 2008 PR blast from Micron, based on their proposed pushing of the 34nm process to its limits. “One Million Write Cycles” was nothing more than a tag line for an achievement accomplished in a lab under ideal conditions. That figure was never reached in anything you could actually buy in a SATA SSD. A better reference would be from that same presentation at the Summit:

flash-2.png

This shows larger process nodes hitting even beyond 1 million cycles (given sufficient additional error bits used for error correction), but remember it has to be something that is available and in a usable capacity to be practical for real world use, and that’s just not the case for the flash in the above chart.

At the end of the day, manufacturers must balance cost, capacity, and longevity. This forces a push towards smaller processes (for more capacity per cost), with the limit being how much endurance they are willing to give up in the process. In the end they choose based on what the customer needs. Enterprise use leans towards SLC or eMLC, as they are willing to spend more for the gain in endurance. Typical PC users get standard MLC and now even TLC, which are *good enough* for that application. It's worth noting that most SSD failures are not due to burning out all of the available flash P/E cycles. The vast majority are due to infant mortality failures of the controller or even due to buggy firmware. I've never written enough to any single consumer SSD (in normal operation) to wear out all of the flash. The closest I've come to a flash-related failure was when I had an ioDrive fail during testing by excessive heat causing a solder pad to lift on one of the flash chips.

All of this said, I’d love to see a revisit to the author’s well-structured paper – only based on the corrected assumptions I’ve outlined above. *That* is the type of paper I would reference when attempting to make *accurate* arguments for SSD endurance.