Leveraging partial-refresh for performance and lifetime improvement of 3D NAND flash memory in cyber-physical systems

  • Jinhua Cui*
  • , Youtao Zhang
  • , Liang Shi
  • , Chun Jason Xue
  • , Jun Yang
  • , Weiguang Liu
  • , Laurence T. Yang
  • *Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

9 Scopus citations

Abstract

Three-dimensional (3D) NAND flash memory-based solid state drives (SSDs) have been widely adopted in cyber-physical systems, due to its performance benefits and scalability. Although 3D flash rapidly increases storage capacity by stacking flash cells in the vertical direction, it faces severe retention errors as well. Periodic refreshing, while effectively mitigating the retention issues, seriously degrades the storage performance and the endurance of 3D NAND flash memory. To address the above challenge, we propose partial-refresh (PR), a novel lightweight data refresh scheme for 3D NAND flash memory in cyber-physical systems. PR leverages LDPC detectability to identify cells that are more vulnerable to errors. By moving these susceptible bits to new pages, PR avoids copying an entire page, and reduces the refresh cost and prolongs the SSD lifetime. Our experimental results show that, on average, a PR-aware flash memory improves refresh performance by 28.2% and extends the SSD lifetime by 4.6% over the state-of-the-art while preserving the high data reliability.

Original languageEnglish
Article number101685
JournalJournal of Systems Architecture
Volume103
DOIs
StatePublished - Feb 2020

Keywords

  • 3D flash memory
  • LDPC
  • Refresh
  • Retention error
  • Write amplification

Fingerprint

Dive into the research topics of 'Leveraging partial-refresh for performance and lifetime improvement of 3D NAND flash memory in cyber-physical systems'. Together they form a unique fingerprint.

Cite this