FTL with NAND FLASH Memory System

25 Jun

NAND FLASH memory access speed because of its light weight, no noise, small size, non-volatile and shock and drop other characteristics have been widely used in high-tech fields and a variety of handheld electronic products. NAND FLASH due to its erase-before-write and erase cycles is limited and other features make the surface used in traditional mechanical magnetic memory file system and applications can not be directly applied to NAND FLASH memory. In order to achieve the two memory compatible for use in NAND FLASH memory system introduced FTL (Flash Translation Layer) layer.

FTL using ectopic updated for every write request is assigned a new, blank position, thus hiding the memory erase-before-write characteristics. Currently a variety of FTL algorithms have been proposed, its main focus is how to assign replacement block write requests, the more commonly used mainly BAST (Block-Associative Sector Translation) algorithm and FAST (Fully-Associative Sector Translation) algorithms. BAST algorithm for each data block assigned one or more replacement block, the garbage collection efficiency is high, but the block utilization is very low; the FAST algorithm is that each replacement block is shared for all data blocks, the block use higher, but garbage collection efficiency is very low. In this paper, ideologically superblock FTL algorithms designed a combined thermal data identified FTL algorithms.

First, the number of adjacent logic block is marked as a block group, the block group as a unit to assign replacement block. During the data update, use the hot data identification algorithm heat each block group access to the system within the block group divided by the weight heat cold, warm and hot categories. For cold data, the system assigns the highest number of blocks has been erased as a replacement block; For hot data, the system assigns the least number of blocks has been erased as a replacement block. Therefore, the thermal properties of good and bad data identification algorithm designed in this paper directly affects the performance of the FTL algorithm. Thermal data identifying the current methods are mostly only focus on LBA (Logical Block Address) address access frequent degree, no dynamic access to frequently reflect the change with time relationship, resulting in a higher rate of error recognition.

This article is designed based on the thermal data queue count recognition algorithm by using the FIFO rule out obsolete data, and give each location are stored in the queue elements to represent different weights at different times of the thermal state of the data; simultaneously flash write with other characteristics, the introduction of the concept of hot zone, so that the required storage space is reduced. Through experiments show that the proposed queue-based count data identification algorithm has a lower thermal error recognition rate can be dynamically reflect LBA address to be accessed frequently change with time relationships. This article is designed to combine thermal data queue count FTL algorithm recognition method of garbage collection and wear leveling efficiency aspects, such as to achieve a better performance.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: