I tried triggering a "Secure Erase" on a 2.5" Kingston A400 SSD in the past, but I did not succeed and instead ended up doing the following with HD Sentinel: one pass of write+read random data; and one pass of write+read zeroes.
Pause for context: My Kingston A400 SSD is used in a Playstation 3, a video game console which was manufactured when 512-bytes sectors and CMR HDDs were the norms. I expect that the SSD will constantly have around 35% of remaining free space, plus additional over-provisioning. (It's a 480 GB SSD, so I guess that the real physical capacity is 512 GB...)
Aside from the additional write operations caused by my fallback method, is it at least a good one for resetting any write amplification "loop" like a real "Secure Erase command" would do?
Is it true that some SSDs should be filled with 1s instead of 0s?
Thanks for anyone having ideas or facts about the topic!
Yes, I can completely understand and agree.
Played / tested lots with the ATA Secure Erase Command but always had troubles with it. This is why exactly Hard Disk Sentinel has no such feature/function.
Not only that this may be not always supported - but even if should be working, many factors (disk controller, controller firmware, drivers, USB/ATA adapters, etc...) can block the command, so it may not be available.
Or if the secure erase procedure can somehow start - then (as the disk drive is not responding during performing the secure erase) a timeout can be triggered: the OS may disconnect the drive completely.
A power cycle/reset/disconnection may cause unkown state at this point.
This is why I completely agree you, personally I do 100% the same and recommend similar for to erase data securely:
1) perform a Disk menu -> Surface test -> Write test with random data (to be sure, to overwrite all sectors)
2) perform a Disk menu -> Surface test -> Write test with zero values (to initialise the sectors and make the drive as empty)
To surely overwrite all sectors, may be good idea to repeat the 1st step twice, to be sure - it depends on how "safe" we want to be and if is there enough time for an additional write cycle.
Alternatively, it is good practice to use the Disk menu -> Surface test -> Reinitialise Disk Surface test - as it performs multiple overwrite passes with different patterns and clearing/verification too.
> Aside from the additional write operations caused by my fallback method, is it at least a good one for resetting any write amplification "loop" like a real "Secure Erase command" would do?
Good question...
According the experiences, the Disk menu -> Extended self test in Hard Disk Sentinel can help in this case after the overwrites - and as a side effect, it can increase the performance too, but it depends on the SSD, firmware etc.
> Is it true that some SSDs should be filled with 1s instead of 0s?
Some SSDs do not actually write zeroes, just marks the actual sectors as empty. You may notice this if the overwrite with random data is performed much slower than overwrite with all zeroes.
But if the drive previously overwritten with any non-zero data (like the random data in step 1 above) then there should be no problem. This may be important only if we want a single pass, quickest possible overwrite: then the simple zero-fill (while fastest) may be not the safest solution.
