In recent years, flash-memory-based Solid-State Drive (SSD) has gradually replaced HDD as main storage device due to
its high performance, low power consumption, and shock resistance natures. The reliability is always a critical issue for Solid-State
Drive (SSD) since each typical flash-memory block can only endure 3,000 to 5,000 erase cycles. Although researchers have proposed
various wear-leveling policies that evenly erase blocks to maximize lifetime of SSDs, the improvement is still limited. The healing
operation, which is based on self-healing architecture, can recover damaged tunnel-oxide layer by heating the cells of flash blocks.
With the healing operation, the endurance of flash-memory blocks can be several times more than that of typical ones. However, its
high latency would largely degrade the performance of SSDs if the healing operation is not properly managed. In this paper, we
propose a latency-aware management scheme, which can hide the overhead of heal latency while taking wear leveling into
consideration. The experimental results showed that our scheme not only achieved better wear leveling (by 72% to 90%) but also
mitigated the impact of healing latency, compared with the related work. With the proposed scheme, the average response time is
improved by 14% to 55% under different traces.

In recent years, flash-memory-based Solid-State Drive (SSD) has gradually replaced HDD as main storage device due to
its high performance, low power consumption, and shock resistance natures. The reliability is always a critical issue for Solid-State
Drive (SSD) since each typical flash-memory block can only endure 3,000 to 5,000 erase cycles. Although researchers have proposed
various wear-leveling policies that evenly erase blocks to maximize lifetime of SSDs, the improvement is still limited. The healing
operation, which is based on self-healing architecture, can recover damaged tunnel-oxide layer by heating the cells of flash blocks.
With the healing operation, the endurance of flash-memory blocks can be several times more than that of typical ones. However, its
high latency would largely degrade the performance of SSDs if the healing operation is not properly managed. In this paper, we
propose a latency-aware management scheme, which can hide the overhead of heal latency while taking wear leveling into
consideration. The experimental results showed that our scheme not only achieved better wear leveling (by 72% to 90%) but also
mitigated the impact of healing latency, compared with the related work. With the proposed scheme, the average response time is
improved by 14% to 55% under different traces.

[1] “Intel md516 nand flash memory js29f16g08aamc1,” 2007.
[2] J.-W. Hsieh, H.-Y. Lin, and D.-L. Yang, “Multi-channel architecture-based
ftl for reliable and high-performance ssd,” IEEE Trans. Comput., vol. 63,
no. 12, pp. 3079–3091, Dec. 2014.
[3] Y.-M. Chang, Y.-H. Chang, J.-J. Chen, T.-W. Kuo, H.-P. Li, and H.-T. Lue,
“On trading wear-leveling with heal-leveling.” San Francisco, CA: Design
Automation Conference (DAC), 2014, pp. 1–6.
[4] L.-P. Chang, “On efficient wear leveling for large-scale flash-memory storage
systems.” New York, NY, USA: Symposium on Applied Computing (SAC),
2007, pp. 1126–1130.
[5] W.-H. Lin and L.-P. Chang, “Dual greedy: Adaptive garbage collection for
page-mapping solid-state disks.” Design, Automation & Test in Europe
Conference & Exhibition (DATE), 2012, pp. 117–122.
[6] Y. Hu, H. Jiang, D. Feng, L. Tian, H. Luo, and C. Ren, “Exploring and
exploiting the multilevel parallelism inside ssds for improved performance
and endurance,” Computers, IEEE Transactions on, pp. 1141–1155, 2013.
[7] S. yeong Park, E. Seo, J.-Y. Shin, S. Maeng, and J. Lee, “Exploiting internal
parallelism of flash-based ssds,” Computers, IEEE Transactions on, pp. 9–12,
2010.
53
[8] Q. Wu, G. Dong, and T. Zhang, “A first study on self-healing solid-state
drives.” Monterey, CA: Memory Workshop (IMW), 2011, pp. 1–4.
[9] H.-T. Lue, P.-Y. Du, C.-P. Chen, W.-C. Chen, C.-C. Hsieh, Y.-H. Hsiao, Y.-
H. Shih, and C.-Y. Lu, “Radically extending the cycling endurance of flash
memory (to >100m cycles) by using built-in thermal annealing to self-heal
the stress-induced damage.” San Francisco, CA: Electron Devices Meeting
(IEDM), 2012, pp. 9.1.1–9.1.4.
[10] R. Chen, Y. Wang, and Z. Shao, “Dheating: Dispersed heating repair for
self-healing nand flash memory.” Hardware/Software Codesign and System
Synthesis (CODES+ISSS), 2013, pp. 1–10.
[11] A. Gupta, Y. Kim, and B. Urgaonkar, “Dftl: a flash translation layer em-
ploying demand-based selective caching of page-level address mappings,” in
ASPLOS, 2009, pp. 229–240.
[12] H. J. J.-S. K. D. Jung, J.-U. Kang and J. Lee., “Superblock ftl: A superblock-
based flash translation layer with a hybrid address translation scheme,” vol. 9.
ACM Transactions on Embedded Computing Systems (TECS), 2010.