Attributes are correct about flash wear—flash wear is not what usually kills an SSD. The controller is what kills the SSD and almost nothing in SMART sees it coming.
Percentage used, reallocated/reserved blocks & uncorrectable errors. These three numbers tell a user enough to get a picture of their SSD's status. The several dozen other attributes mostly tell a normal user very little that is actionable.
CrystalDiskInfo, DriveDx or smartctl should be installed on a different drive. Look for the following four number values rather than the overall verdict on your SSD.
A measure of how much of the SSD's rated write endurance has been consumed. This value will go up from 0 to 100. If your Percentage Used or Media Wearout Indicator reaches 100, that does not mean your SSD has failed. Many users get good use out of their SSDs long after they reach 100% media wearout.
This is the number that represents how much "wear" your SSD has had. This will help in determining if there are any issues with your SSD. For example, a consumer TLC drive is rated for a few hundred terabytes written. A consumer TLC drive would be fine writing 30TB regardless of the percentage.
For NVMe drives. The controller reserves space to use as replacement for bad blocks when they fail. If the available spare begins to fall towards the threshold, it is a legitimate concern; this is the closest an SSD comes to tracking reallocated sectors like a hard drive does.
These are uncorrectable errors from the media that the controller was unable to correct using ECC. If you have any non-zero value on a drive that matters to you then copy your data off immediately. Wear is a measure of how many times you've written to a block, whereas media or data integrity errors are damage rather than consumption.
That's why a perfectly good SSD can go through all tests and just disappear tomorrow.
The majority of SMART information for an SSD does not report anything about the controller. The controller is a small processor running firmware, and if it fails, it will fail entirely without showing any incremental increases in attribute values before failing. It is also the most common form of catastrophic SSD failure.
SSDs have composite temperature reporting. However, many consumer SSDs do not keep track of whether or not they are throttling due to excessive heat. An NVMe drive that keeps hitting its thermal limit under a graphics card may appear normal.
Most consumer level external enclosures do not pass through SMART data from the internal drive. Therefore, the tool will report no SMART data and that should be recognised as being related to the external bridge, rather than the actual drive. On WD units, if you remove the drive to check SMART, you would need to be careful because the encryption key lives on the bridge board.
Consumer SSDs typically do not include power-loss protection capacitors. When there is a sudden loss of power during a write operation, the mapping tables may become corrupted. There is no SMART attribute available that predicts this.
The practical implication for how you should treat each.
Monitoring earns its keep on a hard drive. Reallocated and pending sector counts increase over days or weeks as long as the drive is still readable. That window is where the value lies — you receive warnings, and those are actionable.
An SSD failure that matters typically occurs without the movement of any counters. Therefore "I will back it up when it begins to show problems" is a viable strategy for one and not for another. The only way to reliably protect against failures on solid state storage devices is by having a pre-existing backup, because there is no deterioration period at which you can capture it.
That is something you want to know about before it becomes relevant instead of afterwards. An SSD with 4% wear, but otherwise perfectly healthy attributes is not indicative of safety — it is merely an indicator that nothing measurable has gone wrong yet.
Wear indicators climbing normally. No urgency. Percentage Used increasing steadily as part of a long term write history is simply the drive doing what it was designed for. Replace the drive sometime; don't worry.
Available Spare falling, or integrity errors appearing. Now act. Back up your critical data, most important first, to another physical drive. Don't do a full check/repair — on an SSD these can increase writes/wear without telling you much about problems that have been indicated by the attributes.
If the drive has disappeared from your BIOS or is indicating a completely false capacity then we're into controller territory and not wear. The cost for SSD/NVMe recovery in Manchester is from £300 +VAT after a free 48-hour diagnostic.
Download CrystalDiskInfo, DriveDx, or smartctl onto another drive and view Percentage Used, Available Spare compared to threshold and Media/Data Integrity errors. Ignore the overall 'Good' / 'Caution' score, which is manufacturer defined.
It counts up from 0 as the drive uses up its rated write endurance. The longer you use the drive, the higher the number. This is how the drive works normally; many drives can go far past 100% and still work fine. This is a wear counter. Not a failure predictor.
SMART provides a very accurate way for you to monitor flash wear but doesn't do much when it comes to reporting controller health. Most catastrophic failures come from controllers, so even if all of the SMART attributes look good, there could still be major problems with your drive tomorrow. Also, because no pre-counts exist before failing, a healthy-looking drive could fail completely in the morning.
This represents the spare blocks the NVMe controller has available to swap out any that may fail. A reserve falling towards its threshold is a genuine warning. For an SSD, this is the closest equivalent to a hard drive's reallocated sector count and should be treated as such. Take action now!
No, the lack of SMART support usually means the USB bridge is not passing through SMART information. Be careful about removing the disk to test it, particularly on WD units where the encryption key lives on the bridge board.
No. Deep scanning and using tools to repair issues add writes and wear without providing any new diagnostic information that you didn't already get from looking at the SMART attributes. If you see that some of these values are worsening, make sure to copy your data off instead.