Item

LongliveSSD: Maximizing Lifetime Minimizing Tail-Latency for Multi-Mode SSDs

Ren, Tianyu
Li, Qiao
Lv, Yina
Ye, Min
Guan, Nan
Xue, Chun Jason
Supervisor
Department
Computer Science
Embargo End Date
Type
Conference proceeding
Date
License
Language
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
Driven by escalating big data storage demands and cost-reduction imperatives, NAND flash technology has evolved from single-level cell (SLC) to quad-level cell (QLC). However, QLC NAND flash memory's endurance limitations and suboptimal performance have hindered its enterprise adoption. This paper proposes LongliveSSD to increase the total writable volumes of QLC-based solid-state drives (SSDs). LongliveSSD switches the program mode of blocks level by level (SLC→MLC→TLC→QLC), based on the runtime capac-ity demands. Therefore, each block may experience multiple programming modes throughout its lifetime, dynamically redis-tributing QLC-induced wear to more durable low-density modes for enhanced endurance. Furthermore, we address the long-tail latency challenge of bulk writes for LongliveSSD, through dynamically optimizing programming mode selection based on the request queue patterns. Experimental results demonstrate 2.5× and 1.4x higher improvements in total write volume compared to QLC SSDs and the existing SLC-QLC hybrid SSDs, respectively. Additionally, the performance is improved by 38.3 % and 8.6 %, respectively.
Citation
T. Ren, Q. Li, Y. Lv, M. Ye, N. Guan, C.J. Xue, "LongliveSSD: Maximizing Lifetime Minimizing Tail-Latency for Multi-Mode SSDs," 2026, pp. 30-35.
Source
2025 IEEE 14th Non-Volatile Memory Systems and Applications Symposium (NVMSA)
Conference
IEEE 14th Non-Volatile Memory Systems and Applications Symposium (NVMSA)
Keywords
40 Engineering, 4009 Electronics, Sensors and Digital Hardware, 46 Information and Computing Sciences
Subjects
Source
IEEE 14th Non-Volatile Memory Systems and Applications Symposium (NVMSA)
Publisher
IEEE
Full-text link