본문 바로가기
  • Home

Analysis of the Impact of Buffer Management Policies on Performance in Batch Processing Write Buffers Based on Block Mapping

  • Journal of Internet of Things and Convergence
  • Abbr : JKIOTS
  • 2025, 11(5), 11
  • Publisher : The Korea Internet of Things Society
  • Research Area : Engineering > Computer Science > Internet Information Processing
  • Received : September 9, 2025
  • Accepted : October 10, 2025
  • Published : October 31, 2025

Hyun-Seob Lee 1

1백석대학교

Accredited

ABSTRACT

NAND flash memory-based storage devices have asymmetric data processing units and the physical characteristic of erase before write. Therefore, in order to hide these issues and use them in the same way as general storage devices, a flash transfer layer (FTL) is used to manage logical addresses and physical addresses separately. Recently, performance-optimized page mapping techniques have been widely used. However, this method has the problem of excessive increase in mapping information for mapping logical addresses to physical addresses as the storage device capacity increases. To solve this problem, research is being conducted to reduce the overhead of mapping information. In particular, research on batch processing by collecting data in a write buffer based on block mapping has improved not only the overhead of mapping information but also the performance issues of block mapping techniques. However, further research is needed to optimize the buffer management policy for selecting blocks to be batch processed. Therefore, this study analyzed the impact of buffer replacement algorithms on system performance in a write buffer environment that processes data in blocks. In addition to traditional replacement algorithms such as LRU and FIFO, this study proposes new replacement strategies that consider the characteristics of NAND flash memory and evaluates their performance through experiments. Through experimentation, we analyze the impact of buffer replacement policies on performance and propose research directions for optimizing buffer size and buffer replacement policies.

Citation status

* References for papers published after 2024 are currently being built.