It's the characteristic SATA SSD failure mode that causes the most confusion, as all of your health indicators say that there is plenty of life left in the drive. There are wear reports, but the actual failure is another system.
When an 860 EVO writes data to V-NAND dies, it stores information about where each physical block is located in a mapping table. When this table is lost, the drive can store all of its original content and never be able to access it.
This is why failures occur unexpectedly.
SMART on a SATA SSD will provide several types of data related to flash consumption, such as the Wear Leveling Count, Total LBAs Written, and reallocated block counts. These are valid metrics and the figures were probably correct.
They tell you how much wear has occurred but nothing about the condition of the controller; an always-on little computer running firmware, updating tables for all the logical addresses to their associated physical flash locations, and continually doing garbage collection & wear levelling. If it crashes, it goes down completely, and there are no intermediate readings to show what happened.
A drive at 4% wear could disappear overnight, which is not a contradiction. This behaviour is consistent with the most frequent type of critical SSD failure. As such, backup strategy on solid-state storage cannot rely on detecting degradation.
All are free, and two of them occasionally solve your problem.
Your SATA cable or port could be bad, and the symptoms are virtually indistinguishable from a dead drive. No risk. Only takes two minutes.
A drive that shows up as 1MB, 2MB or 8MB is not dead, and has not been killed by failing memory. It is the controller in a fallback state, because it cannot load its firmware or mapping tables. That being said, the actual data in the flash storage that sits behind it is likely still there.
This will help you determine if the problem lies in the hardware itself (the drive), or if the issue lies somewhere else within your system's architecture (i.e., a faulty controller, driver, or motherboard).
There are three ways a bad drive might act, and each is a different diagnosis. It could either disappear completely, show up at the wrong size, or simply appear as a large block of unrecognisable junk. Power cycling repeatedly does absolutely no good and could potentially lead to even more writing on top of what already exists.
It's worth pointing out clearly as manufacturers' tooling seems to encourage it.
An apparent solution when you have a malfunctioning drive is an update to its firmware via the utility provided by the manufacturer. However, if the drive is already in a "fallback" mode, this action is the wrong move.
Updating firmware will write over the service area which is where the recovery mechanism must access the mapping tables. If the service area is already bad, then the firmware update can stop partway through, leaving your drive in a worse and possibly permanent state. This also applies to secure erase, and any other 'fix' or 'restore' options from the manufacturer.
You should keep your firmware up-to-date with all drives so long as they are functioning properly and consider it preventive maintenance. Do not reach for it as a fix once a drive has started behaving strangely.
The flash memory is still intact, behind a layer that no longer responds.
The manufacturers of computer parts include a diagnostic function in their controllers that bypasses the normal user interface. In this way, the service area can be accessed by reading it directly, including the translation tables used to translate the logical address to the physical location of the flash memory. These tables are all that is required to perform recovery since if they do not exist then the data contained within the NAND flash memory is nothing but unorganised data, due to wear levelling which causes the physical organisation of the NAND flash to be completely unrelated to its logical organisation.
When the controller itself cannot be recovered, in some cases the flash memory may be able to be accessed directly, de-interleaved and corrected using ECC. This method is much slower and more risky than attempting to recover the controller, and on many modern devices will return data the controller had encrypted (a complete duplicate of something that is unreadable). For this reason, recovering the controller is always attempted before accessing the flash memory.
There are two ways in which the recovery process can be terminated: TRIM (which physically erases deleted blocks within minutes on a powered drive), and encryption without a password. Both are established at the free diagnostic. SSD and NVMe recovery in Manchester is from £300 +VAT, with a 50% deposit for controller-level work.
Typically, it's the controller rather than the NAND flash memory. Although the NAND flash memory contains your data, it is usually intact behind layers that stopped responding. It can frequently be accessed via the manufacturer's diagnostic mode and then read out non-destructively.
The reason for this problem is that SMART measures flash wear, but does not measure the health of the controller. The wear figures were probably accurate; they are describing an entirely different issue than what ultimately caused the device to stop functioning.
No. An upgrade will rewrite the service area (the very area that the recovery operation must be able to access). On a drive already in a fallback state, the firmware update may only partially succeed and make the condition permanent. This also applies to secure erase.
The controller is in a deliberate fallback state. The flash hasn't disappeared. There isn't a single drive made anywhere that size – it's a placeholder used by the controller if it can't load its firmware or mapping tables.
It rarely comes from an active working drive, because TRIM will erase those blocks of data physically within minutes. A drive that has already failed can't use TRIM, so a failure sometimes results in saving what continuing health would destroy.
The cost will be from £300 + VAT after a free 48-hour diagnostic has been done. There will be a 50% deposit required before we do any controller level work. When we run the diagnostics on your machine we can tell at that point if the data is recoverable; when it is TRIMmed or encrypted (without a key) – we'll let you know at that time, rather than after we have issued an invoice.