Flash memory has many benefits over conventional rotating media including compact size, low power consumption, zero seek time, and better shock resistance. These properties make flash memory storage ideal for the mobile environment including digital cameras, portable devices, and other embedded systems. However, NAND flash has a limited number of write/erase cycles. There are many sophisticated algorithms for evening out the usage of flash blocks across the whole address space (wear leveling) to improve the expected life of a product. In addition, NAND devices contain a certain number of bad blocks and require extensive error correction to be performed on the read data. A host system needs to consider handling and managing bad blocks, not just the ones identified at initialization, but also the ones that go bad when in use. Next, most NAND devices require error-correction code (ECC) for the same data block size. Implementing an error-correction algorithm in firmware will consume lots of precious arithmetical computations, and if the error frequency gets higher, it will slow the system performance down significantly as well as lower the overall throughput. Moreover, the feature of power loss recovery should demonstrate what is it all about during sudden power interruptions or brown-out conditions and where the inadvertent removal of a battery could threaten the integrity of data. Overall, how a storage system manages the memory is the key to understanding the extended reliability of the host that relies on these storage systems.
If you like to download the white paper, please Contact Us.