Google said in its recent report that bad block has been witnessed at a 30% to 80% chance in four years of use of solid state drives (SSDs) for servers at data centers. Bad block can be defined as a phenomenon in which a physical or electrical damage in a memory storage system makes certain memory cells irrecoverable.
SSDs in general have dozens of NAND flash memories, controller chips, etc. The former store data while the latter record and retrieve certain data. According to experts, durability deteriorates and bad block occurs when reading and writing are repeated thousands of times in an MLC NAND flash memory in wide use. Then, cells subject to the bad block lose their data. According to Google, an SSD storage system is normally less error-prone than a hard disk drive (HDD) storage system but has a higher uncorrectable bit error rate. It added that this problem is more frequent in NAND products that have a floating gate structure.
In the meantime, Samsung Electronics, which is the largest NAND supplier in the world, recently adopted charge trap flash (CTF) for 3D NAND production in order to deal with this SSD NAND reliability issue. CTF is a technique for keeping an electrical signal in a non-conductor and can ensure a higher level of stability than the floating gate structure.
“SSDs can be used as a main storage medium in the server market but its reliability has not reached a level at which HDDs can be completely replaced,” said Do Hyun-woo, research analyst at Mirae Asset Securities, adding, “It seems that HDDs will continue to be used as a main storage medium even after SSDs become prevalent.”