Abstract
When applying the existing flash translation layer technique to a mixed NAND flash storage device composed of Quad Level Cell and Single Level Cell, because the characteristics of a semiconductor chip are not taken into consideration, the data are stored indiscriminately, and thus the performance and stability are not guaranteed. Therefore, this study proposes a flash translation layer algorithm using the warm block technique in a NAND flash storage device that combines a large capacity Quad Level Cell and a high performance Single Level Cell. The warm block technique avoids overloading of the read/write/erase operations in the Quad Level Cell flash memory by efficiently placing hot data that are frequently updated on a long-living Single Level Cell. It was confirmed experimentally that the lifetime extension and performance of hybrid NAND flash memory are improved using the warm block technique.
Highlights
Flash memory is classified into various types of semiconductor chips including Single Level Cell (SLC), Multi- Level Cell (MLC), Triple Level Cell (TLC), and Quad Level Cell (QLC) depending on the number of bits that can be stored in a single memory cell
The present study proposes flash translation layer (FTL) algorithm based on the warm block technique in a NAND flash storage system that combines a high-capacity QLC and a high-performance SLC when considering the characteristics of a semiconductor chip
We proposed an algorithm for minimizing the erase operations of a QLC semiconductor chip and increasing the input/output performance of an SLC semiconductor chip by utilizing the warm block technique in the mass storage system, which is a mixture of QLC and SLC
Summary
Flash memory is classified into various types of semiconductor chips including Single Level Cell (SLC), Multi- Level Cell (MLC), Triple Level Cell (TLC), and Quad Level Cell (QLC) depending on the number of bits that can be stored in a single memory cell. A hybrid SSD, in which various types of chips are mixed and used in a single storage device, has recently been proposed [1] Because such a storage device has high-grade and intermediate/entry-level semiconductor chips in a single SSD, it is necessary to decide where to store the data, namely, in either the high-grade or intermediate/low-cost semiconductor chips when a write request is made from the file system. Most existing approaches [3] write updates to the log block When such a log block-based FTL is applied to hybrid NAND flash memory [4], the characteristics of each semiconductor chip are not considered and data transfers between the high-grade and intermediate/low-cost semiconductor chips frequently occur, resulting in a degraded performance. The algorithms in this paper are designed to minimize the erase operation of intermediate/supplied semiconductor chips and increase I/O performance of high-end semiconductor chips, which are for large storage devices based on QLC+SLC mixed NAND flash.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
