SSD 내구성 강화를 위한 Window 기반 오프라인 Hot/Cold 데이터 분류 기법

초록

NAND flash memory-based storage devices, such as solid-state drives (SSDs), use a software layer called the flash translation layer (FTL) to function as block storage devices. The FTL provides various functions, including data placement policies, address mapping, wear leveling, and garbage collection, all of which significantly impact the performance of the storage device. Accordingly, various studies have proposed techniques to improve the performance of each FTL function. This paper proposes an offline data placement policy to minimize write amplification in NAND flash memory. The proposed data placement policy analyzes trace data to distinguish between hot and cold pages based on the likelihood of future re-access and adjusts the window size to perform efficient data placement. Additionally, the trace is divided into chunks, and the optimal window size for each chunk is determined. Thus, the write amplification factor is kept close to 1 to produce optimal data placement results under the constraints of limited resources and time.

키워드