Power loss SSD drives may lead to data loss

I've been a proponent for several years of SSD's. In recently examining a 500 GB Crucial drive offering (on sale for $58; https://www.amazon.com …ND-Internal/dp/B0784SLQM6) I noticed the comment "Integrated Power Loss Immunity preserves all your saved work if the power unexpectedly gets cut". That started me investigating as this seemed a concern that I never considered. We discuss from time to time the need for backups but it seems like this should be a big concern for all of us as well and we need to consider not doing work during high risk times (thunderstorms) and verify the proper operation of any UPS that we might use. I happen to use a UPS on my main working PC and a quick test indicates it works, perhaps in spite of a three year old battery!

The first content discussing the subject is at https://www.extremetec​h.com …outages-killing-your-ssds. Although the article is from 2013, apparently back then, as evidenced with 15 different brands of drives, a power loss could result in varying degrees of data corruption.

Another good article, "Risks of unexpected power loss on solid state drives" by Hewlett Packard can be found at http://www8.hp.com …PDF.aspx/4AA6-1470ENW.pdf.

John from PA wrote in post #18797686
I've been a proponent for several years of SSD's. In recently examining a 500 GB Crucial drive offering (on sale for $58; https://www.amazon.com …ND-Internal/dp/B0784SLQM6) I noticed the comment "Integrated Power Loss Immunity preserves all your saved work if the power unexpectedly gets cut". That started me investigating as this seemed a concern that I never considered. We discuss from time to time the need for backups but it seems like this should be a big concern for all of us as well and we need to consider not doing work during high risk times (thunderstorms) and verify the proper operation of any UPS that we might use. I happen to use a UPS on my main working PC and a quick test indicates it works, perhaps in spite of a three year old battery!

The first content discussing the subject is at https://www.extremetec​h.com …outages-killing-your-ssds. Although the article is from 2013, apparently back then, as evidenced with 15 different brands of drives, a power loss could result in varying degrees of data corruption.

Another good article, "Risks of unexpected power loss on solid state drives" by Hewlett Packard can be found at http://www8.hp.com …PDF.aspx/4AA6-1470ENW.pdf.

Good information. But the same sort of thing can happen with the SD cards we use, or the USB drives, especially if a file-write operation is occurring. That's why it is best to use whatever procedure for removing a drive from your computer and wait until the machine indicates it is safe to remove a drive. This ensures files have been closed properly.

Capn Jack wrote in post #18797841
Good information. But the same sort of thing can happen with the SD cards we use, or the USB drives, especially if a file-write operation is occurring. That's why it is best to use whatever procedure for removing a drive from your computer and wait until the machine indicates it is safe to remove a drive. This ensures files have been closed properly.

If you run Windows, the safest thing is to disable disk write caching. That way you don't have to wait for the safe to remove message. You can learn how to disable the feature here...https://support.micros​oft.com …k-write-caching-on-or-off

That prevents user stupidity from preventing the Write from completing in the middle of that process, but it does nothing for power interruption to desktop PC when there is an electrical interruption of service due to storm or car taking down a pole.

John from PA wrote in post #18797899
If you run Windows, the safest thing is to disable disk write caching. That way you don't have to wait for the safe to remove message. You can learn out to disbale the feature here...https://support.micros​oft.com …k-write-caching-on-or-off

Thanks, but my preference is to let the machine let me know it is OK to remove the drive.
Nothing wrong with your suggestion, but a couple of times I got a reminder I hadn't finished working on a file and needed to close it properly. The trade-off is the original reason for this thread- data can be damaged if power is lost, but I'm running on a UPS or a laptop (essentially contains a UPS) so I don't have to worry about power issues.

Capn Jack wrote in post #18797901
Thanks, but my preference is to let the machine let me know it is OK to remove the drive.

That's true, but I have had a few occasions where the computer wouldn't let me eject a USB drive even though there was no write process that I had initiated. The only solution I could think of was to shut down the computer and remove the drive once shut down. Wouldn't this alleviate that, so long as you verified that any writing you had initiated was completed before ejecting?

el_duderino04 wrote in post #18798398
That's true, but I have had a few occasions where the computer wouldn't let me eject a USB drive even though there was no write process that I had initiated. The only solution I could think of was to shut down the computer and remove the drive once shut down. Wouldn't this alleviate that, so long as you verified that any writing you had initiated was completed before ejecting?

Sorry, I don't know.
It depends on why the computer "thought" a file was still open. For example, the warning would occur if a program bug failed to close a file properly. It would probably be safe to yank the drive using either method in this thread, in that case. I know what you are describing, and it has happened to me, although not in recent memory. It seemed to correlate with sending an attachment via Outlook from a USB drive and I handled it the same as you described.

I move data from Linux systems onto Windows using USB drives; Win10 always claims the drive is damaged and wants to fix it. I let it do so, and I can then access the files. So Windows seems to be picky about something.

The sudden power lost in flash storage has been and still is one of the big consideration in the design. The industry call this UGSD (Ungraceful shut down) or WA (write abort). There are multiple issues associated with this and there are too many solution in the industry, some work better than the other. This so called power lost protection is usually NOT in USB drive at all, but SD sure have it (more basic version) and SSD will have very fancy schemes to handle this this. So it is safe to say that any SSDs in the last 8-10 years will have some kind of power lost protection. Are they 100% power lost proof, definitley not, but as technique improve, the modern schemes are much more robust in general.

When I say protection, I don't mean the file that is being written at the time of power lost. That few option can protect, usually expansive. Most of client level SSDs are decided such that if power lost happen, the data completely written before the power lost is safe. And the data to be written after the power lost is also safe.

eelnoraa wrote in post #18811639
The sudden power lost in flash storage has been and still is one of the big consideration in the design. The industry call this UGSD (Ungraceful shut down) or WA (write abort). There are multiple issues associated with this and there are too many solution in the industry, some work better than the other. This so called power lost protection is usually NOT in USB drive at all, but SD sure have it (more basic version) and SSD will have very fancy schemes to handle this this. So it is safe to say that any SSDs in the last 8-10 years will have some kind of power lost protection. Are they 100% power lost proof, definitley not, but as technique improve, the modern schemes are much more robust in general.

When I say protection, I don't mean the file that is being written at the time of power lost. That few option can protect, usually expansive. Most of client level SSDs are decided such that if power lost happen, the data completely written before the power lost is safe. And the data to be written after the power lost is also safe.

WIth the conventional harddrive, the data already written is safe, too. And if, in the middle of writing data to the harddrive, power is lost, the heads are immediately withdrawn from the proximity of the platters so that the air cushion does not go away and allow the heads to crash into the still spinning (momentum) platters. So a mid-write power loss to SSD is not unlike in the HD, unless there is some design consideration in the SSD which makes it somewhat less susceptible to mid-write data loss than the HD.

With a non-battery computer (e.g. desktop) the sudden power interruption of AC would seem to be similar regardless of data storage technology involved, unless a very tiny file is being written to SSD.

Wilt wrote in post #18811717
WIth the conventional harddrive, the data already written is safe, too. And if, in the middle of writing data to the harddrive, power is lost, the heads are immediately withdrawn from the proximity of the platters so that the air cushion does not go away and allow the heads to crash into the still spinning (momentum) platters. So a mid-write power loss to SSD is not unlike in the HD, unless there is some design consideration in the SSD which makes it somewhat less susceptible to mid-write data loss than the HD.

I believe what you say is true with respect to a singular loss of power; someone tripping over a power cord for instance. However, It is my understanding that a "flickering" type of power disruption that might be encountered in a thunderstorm can cause damage to a hard drive. Since it is a function of the head parking process, I'm not sure what damage might occur to an SSD under similar circumstances. Perhaps "eelnoraa" can shed some light on that type of event.

I just found this information in the context of power disruption affects to a harddrive

"A summer thunderstorm that leaves you in the dark for even just a few minutes can damage your computer or external hard drive. Faulty hardware can also cause power surges that can short out a data storage device. There are many ways a data storage device can fail as a result of a blackout or power surge. A sudden loss or spike of power can short out your hard drive’s control board. It could cause your hard drive’s read/write heads to crash, its motor to seize, or its platters to become damaged. ...

"a sudden loss of power might cause corrupted data to be written to your hard drive’s boot sector or critical operating system files. This can cause your PC to have trouble booting up. It’s also rare, but possible for a sudden loss of power to cause your PC’s hard drive to break down.

"A sudden loss of power can cause many different kinds of failure in a hard disk drive. Hard drives have several moving parts. All of these moving parts are potential points of failure. When you power down your computer, it sends signals to your hard drive. These signals tell the hard drive to prepare itself for shutdown. The read/write heads unpark from their position over the platters and the platters slowly spin down.

When your computer or external device abruptly loses power, the hard drive doesn’t receive these signals. The flow of power through the PCB to the hard drive spindle motor stops without warning. The heads might not have time to unpark before the cushion of air keeping them afloat above the platters dissipates.

If the hard drive is in the middle of a write operation, data can become corrupted. If a firmware sector becomes corrupted, the hard drive can be prevented from booting. The read/write can also make physical contact with the platters, damaging both.

If the heads make contact with the platters, they can stop them from spinning. This harms not only the heads and platters, but the motor as well. The hard drive spindle motor can become seized if it encounters sudden resistance.

If the platters keep spinning, the read/write heads will start to gouge out the magnetic coating on the surfaces of the platters. This is called rotational scoring. Severe rotational scoring can render the data on a hard drive unsalvageable."

Let me give a simple explanation why power lost can be an issue for NAND flash. The problem is not the power lost itself, it is when the power come back on, how accurate the drive system find the last "written" or better described as write aborted location. If this location is found correctly and then handle correctly, no data will be lost. But if this location is NOT found correctly, let's say the write aborted page is mis-detected as "erase" state. The drive think it is OK to start writing new data at this location. Two of the most likely scenarios are these:

1. FW write new data on top of previous "half written" data. This almost guarantees the new data will be corrupted beyond ECC correctable. The reason is NAND flash can't over write previous data, they accumulate program states. In modern/higher end SSD, say WD blue or black, there is also XOR parity protection. Single page un-correctable can usually be recovered by XOR parity. But for lower end/budget SSD, if this feature isn't implemented, data will be lost.

2. When writing new data on top of previous "half written" data, there is a chance that the page will be over-programmed. In this case, all pages in the block will not corrupted. This is usually referred as "pinch off". Here, due to multiple pages become correctable, even XOR parity won't help.

Sudden power lost, or Ungraceful Shutdown, is a big deal in NAND flash storage design. These are just two simple example. I would say during an SSD design, 50% of the corner cases/bugs are somehow related to sudden power lost. Manufactures spend a lot of time and efforts to sort this out

This is why it's always a good idea to eject your media!

el_duderino04 wrote in post #18798398
That's true, but I have had a few occasions where the computer wouldn't let me eject a USB drive even though there was no write process that I had initiated. The only solution I could think of was to shut down the computer and remove the drive once shut down. Wouldn't this alleviate that, so long as you verified that any writing you had initiated was completed before ejecting?

I found a way this happens in Windows 10. If you have "file preview" turned on (Menu item View, Preview pane), and a file is selected, the system says a file is in use when ejecting a drive. Closing the window containing the file allows the drive to be ejected with no message. Whatever allows the preview has the file opened.

John from PA wrote in post #18797686
I've been a proponent for several years of SSD's. In recently examining a 500 GB Crucial drive offering (on sale for $58; https://www.amazon.com …ND-Internal/dp/B0784SLQM6) I noticed the comment "Integrated Power Loss Immunity preserves all your saved work if the power unexpectedly gets cut". That started me investigating as this seemed a concern that I never considered. We discuss from time to time the need for backups but it seems like this should be a big concern for all of us as well and we need to consider not doing work during high risk times (thunderstorms) and verify the proper operation of any UPS that we might use. I happen to use a UPS on my main working PC and a quick test indicates it works, perhaps in spite of a three year old battery!

The first content discussing the subject is at https://www.extremetec​h.com …outages-killing-your-ssds. Although the article is from 2013, apparently back then, as evidenced with 15 different brands of drives, a power loss could result in varying degrees of data corruption.

Another good article, "Risks of unexpected power loss on solid state drives" by Hewlett Packard can be found at http://www8.hp.com …PDF.aspx/4AA6-1470ENW.pdf.

Interesting read John.
I will get a chance later today to look at it.
Since we live in the same area and we do get our fair share of storms I learned a very simple solution. As our old business was supplying the Physical Therapy industry as a manufacturers rep. Our main company we dealt with was in Tulsa OK. They were manufacturers of Ultrasound and Electrical Stimulation devises for the rehabilitation market. Many times we would call there and there computers and other valuables were shut down. When I finally asked the service tech about what they use for surge protection during their severe weather situation, he told me the worlds best surge protection is an unplugged AC Power Supply.
To this day, when Mother Nature rears her ugly head, I still reach underneath and follow his lead. I just bought an APC UPS 1500 for both here in Wallingford and in our home in State College. For now I do find myself leaving my iMac 27 plugged in while not in use. Come the summer I may just do what I did before to be safe.
Thanks for the info John.