Test your SSDs or risk massive data loss, researchers warn

I agree, otherwise this test means almost nothing IMO.

yes, i also have that question, what kind of software are you guys using?
MHDD, HDAT, etc. or others?

 

Hahahaha poor bill.

 

I had the benefit of downloading and reading through the report PDF!

Cheers!

 

Well, I did too. I was just hasty interpreting what I read and (I think) incorrectly took it as though they subjected each SSD to 3000 outages, and not that 15 were subjected to 3000 total.

And while 200 unexpected outages during the lifetime of SSDs may be closer to reality for power outages, subjecting computer hardware - especially in a data center, is not.

It is important to note that 1, just a single unexpected power outage can result in data loss and/or drive corruption - depending on where during a "write" cycle the outage occurs.

It is also important to note that many "real-world" unexpected outages are not "clean". That is, power often "flickers" momentarily before total outages and often again before a "stable" return. And the sudden voltage rise phases of those power fluctuations often consist of huge spikes and surges which bang hard on our power supplies, severely taxing the PSU and motherboard regulator circuits - which, if nothing else, ages those components more so than stable power.

Again - every computer should be on a "good" UPS with AVR.

Also note that power "flickers" are long-duration outages - at least when it comes to high speed digital electronics. If an outage is long enough for slow human perception to detect a flicker, it might as well be an eternity of abuse to the CPU and other sensitive devices.

Note my 1500VA (~950W) APC UPS with AVR easily supports my i7, hungry graphics card and 8Gb of RAM computer, all my network devices, my phone, AND my two 22" widescreen monitors. And note too, backup power support during power outages is only the icing on the cake. The bread and butter of a "good" UPS with AVR is the automatic voltage regulation it provides 24/7/365. That said, my UPS will provide at least 20 minutes of backup power during a full outage - way more time than I need to "gracefully" close all my open files, exit the OS, and shutdown the computer (all of which can be done automatically via Windows, or the UPS maker's software - if the UPS has a communications link with the computer).

Remember, a surge and spike protector is little more than a fancy and expensive extension cord that will not prevent a sudden shutdown. In fact, they protect your hardware (while hopefully preserving your data) by cutting power suddenly instead of letting harmful spikes and surges through. That's fine if all you are concerned about is your hardware. But for many computer users, including your's truly, (1) the data on my computer is much more valuable than the hardware and (2) computer uptime is a mission critical necessity.

 

Agree with your views regarding surge protectors vis a vis UPS – unfortunately a decent UPS is a very expensive item for home users. I have had hardware 'saved' by one of these cheap, single use power surge protectors. Whether that would have made a difference had I had an SSD is unknown.

I would have preferred this research team to have decreased the number of fault cycles to something more realistic – is 200 power failures realistic? It would not seem so to me.
What is a meaningful 'life' for a hard drive? At say, 5 years – this would represent at least 3 power failures per month – every month over the 5 years! Hardly realistic. Even at an 8 year life – that's still 2 per month! (does anyone keep a hard drive for 8 years?)

I think that the report has rather different conotations for enterprise users as compared to home users. The work loads of Data Centres, in particular, are vastly different and therefore liable to far more potential damage should a power failure occur. But, like you say, such centres should have some form of emergency 'protection', whether that be a UPS or their own power generators which kick in at such times.
For home users – any doubts regarding the effects of power failures should be explored further so that positive recommendations can be provided. Eg, ''To protect your SSD you must do this – or you must do that'' – etc.

(Meanwhile – my battery pack stays in! It's now mid summer – and Power authorities in this country have a nasty habit of unexpectedly cutting the power to whole suburbs – they call it 'load shedding' – Duuh!)

 

You don't need to buy a $400 UPS but you should avoid a $50 one. And yes, they are expensive considering they are devices you really hope you never actually need.

Nevertheless, $400 is much cheaper than replacing all your hardware, plus the manhours, cost of downtime, and frustration created if you lose everything.

 

If you intend to use a generator to back up your UPS, make sure to get an inline UPS. It converts line power to sine wave output. Computers may not accept square(ish) wave power from consumer generators.

Yes!

Also, make sure to configure your setup so the computer shuts down properly before the UPS runs dry.

 

i have 6 SSD's in my house, 4 are first Gen SSD's. i must have experienced around 3 power outages (whole street with no power) and around 6 induced power outages (caused by me via circuit breaker). no data on any SSD was affected so from my experience, if there is a problem with SSD's when it comes to power outages then it does not seem to be any better or worse than hard drives.

plus if you aint backing up, power outages will be one of the last things to be concerned about as you are more likely to be affected by user error/viruses etc to be responsible for destroying your data.

note- never owned a UPS

 

Does turning you PC off manually with the power button also count as unexpected interruption of power or has it enough time to stop reading/writing operations?

I guess the necessity of a UPS is very dependent on the place you live. I hear there is a lot more 'dirty' electricity(power fluctuations/spikes) and power outages in countries where the power lines are above the ground. I live in the Netherlands where it is largely under the ground and can't even remember the last power outage, that happens perhaps just a few times in a lifetime.

 

good question, i would like to know this too, although i suspect not.

i recently went on a holiday to Amsterdam, lovely place, my first time at the cafe's . the wife's first time too.

i forgot to mention in my previous post that although i do not use a UPS, i do use an extension lead with power surge protection. the electricity in England is very stable so not sure how much times the surge protector has helped me

 

That depends on how the power button's operation is set in the BIOS Setup Menu.

If the button is set to "shut down" the system, then typically it will "gracefully" close the opened Windows files, and then turn off the computer (put it in stand-by mode actually). But it may not properly save other opened files like Word or email documents.

In any event, this is easy to determine by watching your monitor.

 

Some of us are living in the Third World

 

Without adopting an "I told you so" attitude, this illustrates a shortcoming of SSD technology. And it's an artificially created one too.

It would cost next to nothing to design in a small SuperCapacitor that maintains power to the SSD for perhaps 5 or 10 seconds regardless of what's going on with the host system. A mini UPS built right in.

This isn't rocket science. No major (or minor) advances in tech need to happen. No new materials need to be invented. No killer software.

These extra "lifeline" seconds would be enough flush any cache and wrap up any operations currently in progress. It would enable an orderly shutdown in case of power failure. It could potentially lose the work in progress file because the host system stopped sending the data mid-stream. But the already existing SSD contents would remain uncorrupted.

SSD's can do so much more than spinners when it comes to self-preservation and maintenance.

But the industry is so intent on cost-cutting it can't even implement this simple function. A SuperCapacitor would add perhaps $10 to the cost of the SSD. But competition is too intense to permit such a price increase.

 

While your capacitor idea could technically work, I do not believe it is practical or cost effective, and it presents some safety issues too. And there just is not the need for such a dedicated, single purpose feature either.

For one, much of the problem during a sudden power loss is due to the fact the CPU stops, as does motherboard bus traffic. When the CPU and buses stop processing data, so does the OS and therefore, for the SSD to keep working, it would have to include a lot of intelligence, and a large buffer to temporarily hold the data, as well as the internal power source (cap, battery, or something) to keep them going.

Then it would need to ensure voltage is totally isolated from potential external exposure by a less than careful user. The worry is not injury to the user, but a destructive short damaging the SSD, or the motherboard's interface, or both.

I believe all that would add considerably more than $10. And since there are dozens of SSDs for under $100, adding 10%, 15%, or even 20% to the cost for something not needed by the vast majority of users just ain't good business.

The better, cheaper, more robust solution is to (1) ensure good facility power, and (2) have current backups. Ensuring facility power can be difficult for many users, so again, get a "good" UPS with AVR which will protect your entire computer, to include your SSDs, all your network gear, and your monitors too (if LCD), and still have 20+ minutes of battery runtime.

 

I believe that battery modules for my RAID controllers cost more than 100 USD!

 

There is no option for that in my BIOS, if I click the power button once it will go in standby and if I hold it down for a few seconds it will turn off immediately.

 

That is the typical default settings. It is to prevent folks from losing all their data when little Russel pushes the button, then asks what it does.

 

Well, it is indeed nice, but the question is whether turning it off like that(with the power button held down) counts as unexpected power interruption. If you need to turn it off like that because your computer freezes or if you want to boot from cd/usb and it doesn't recognize it and you want to try again, clean the disk etc. If that would also count, then the average number of unexpected power interruptions would be a lot higher I think.

 

Now you are talking about two different things. If your computer is locked up, you have no choice. You have to kill the power to reboot. And yes, if the OS has not loaded, that counts as an unexpected power interruption.

But if your computer is not locked up, you should always "gracefully" exit your operating system, then shutdown the computer.

Do not confuse "Standby mode" with "Off" or with one of the "Sleep Modes". When you use your front panel button to turn the computer off, unless you unplug from the wall, you flip the "master power" button on the back of your PSU (if it has one), your computer is in "Stand by" and not "Off".

The ATX Form Factor standard calls for all PSUs to supply +5Vsb standby voltage to several points on the motherboard when computer is turned off, but still plugged in (or switch in back set to on). This allows for various "Wake on..." commands like Wake on Keyboard, Wake on Mouse, or Wake on Power Button.

So when you say if you simply press and release it goes into standby, one of two things is happening. Either (1) it goes to sleep and when awakened, all your applications and documents are already opened right where you left off, or it (2) goes into Standby mode where all your applications are closed and any open documents are automatically saved, then when you start again, you are at your blank desktop.

If it was in Standby, you can safely kill the power. If it was in one of the sleep modes, you should wake it up, then gracefully exit Windows and shutdown before killing power.

 

All I can say from MY own experience, your's may differ: no problems at all with my own SSD (Vertex 2) or it's data integrity (that I know of) or whatsoever after turning off power this way a few times due to a frozen software. Same with a relative (also Vertex 2) who does this more often than I do. Bottom line: those SSD's are working very reliable so far and I have really no concern for the future. Use backups if you fear disaster, but use SSD *NOW* and don't wait another 3 years (because of unnecessary fear!) before leaving slow OS-HDD behind ... you will smile all the time because it's so fast.

Or you could worry all the time (day and night) about the next meteroid passing earth, sure!

 

SSD have become decently priced starting only a few months ago.

a few years back i decided i would not get involved with them unless the prices dropped below 1$ per GB.

which it is now and still going down in prices.

very glad i got one a couple months ago.
very much well worth the bang for the buck.

 

I am planning to buy a SSD but the factor delaying that is not concern of a faulty SSD, but lack of money

I forgot to mention that I have a notebook, not a desktop. If I press the power button once, it goes into some sort of standby that seems the same as Windows' standby and the power led is blinking orange, so it's not off. I know now that turning it off forcefully counts as unexpected, but I'm still not sure if turning it off while still in the BIOS part of booting also counts as unexpected. When it is still in that early stage it isn't necessary to hold the button down as it powers off immediately.

 

This sums it up pretty nicely:

Essentially the way they cut the power is designed to circumvent all safety features enterprise grade SSDs or disk controllers have. This makes the entire paper kind of misleading and pointless. What will happen next? Disastrous car crash tests with cut seat belts and disabled airbags?

 

How would those "safety features" come into play if a power cable or UPS failed? That's what they were simulating, I think.

Is is really true that SSDs (even enterprise SSDs) don't have backup batteries to protect cache? That's standard for enterprise RAID controllers.

 

I haven't seen battery backups on SSDs yet. My guess is because battery backups must be replaced from time to time and they can be rather large. Most recent enterprise level SSDs and even some consumer level SSDs come with supercapacitors though that should allow the disk to shut down properly in case the power is cut.