Michaelab opened this issue on May 10, 2013 ยท 14 posts
Dale B posted Fri, 10 May 2013 at 5:53 PM
The problem with SSD's goes down to the atomic level. Each and every write event does damage to the substrate at the atomic level; a good analogy is like shooting a 12'x12' sheet of thick paper with a BB gun...only the paper changes color every time it is hit. That's a write state. You can shoot so many holes in the sheet then its structure becomes so fragile it collapses. Atomicity is the rule for flashRAM, which is what SSD's are (ever wondered why a USB stick drive stops working when you have treated it like glass? This is why). You only have X number of write cycles to play with before you get a junction failure, and you lose one transistor in an SSD, the byte it was involved with is gone. Unless you go with a custom designed drive, the -only- way to recover data from an SSD is to open the shell, find and remove the bad flash chip, then access the remaining chips. And you can only do that if they do not have a RAID configuration internally. Which nigh onto all of them do. It's the only way to get the r/w speeds up to useable levels (and the internal configuration is almost always RAID 0, or data striping. Lots faster that way....but if any chip fails, you have lost all data in the drive).