在裝置管理的疊瓦式硬碟(DM-SMR)中，當資料更新或是當寫入資料會造成相鄰磁軌的資料受損時，會使用一部分硬碟空間作為持久性快取(PC)來暫存這些寫入資料。當持久性快取的空間不足時，疊瓦式硬碟會執行空間回收來增加持久性快取的可用空間。疊瓦式硬碟的空間回收需要很長的時間來完成，若是在I/O請求執行中途觸發了空間回收，會導致後續I/O請求的延遲，是造成疊瓦式硬碟效能降低的主要原因。因此我們提出一個方法透過在不同長度的空閒週期中執行適當的空間回收方法來降低疊瓦式硬碟的長尾延遲進而提升效能。實驗結果顯示，我們的方法跟目前最新的Skylight、KFR、Idler 和 RL Cleanig相比，分別降低了75%、34%、36%及53%的長尾延遲。

DM-SMR disks use a portion of disk space as persistent cache (PC) to address the overlapping issue when updating data in the overlapping tracks will cause damage to the data stored in the adjacent tracks. When the PC space is insufficient, a PC cleaning will be triggered to reclaim its invalid space but the PC cleaning is time consuming and will cause the long-tail latency of DM-SMR disks. In the paper, we will propose a space cleaning method with different idle periods to reduce the long-tail latency of DM-SMR disks. The objective is to perform a proper space cleaning in a suitable idle period to avoid delaying the subsequent I/O requests and also reduce the long-tail latency of DM-SMR disks. The experimental results show that the proposed method can reduce the long-tail latency of DM-SMR disks by up to 75\%, 34\%, 36\%, 53\% on average when compared to Skylight, KFR, Idler, and RL Cleaning respectively.

[1] S. Greaves, Y. Kanai, and H. Muraoka, “Shingled recording for 2–3 tbit/in2,” IEEE Transactions on Magnetics, vol. 45, no. 10, pp. 3823–3829, 2009.
[2] Y. Kanai, Y. Jinbo, T. Tsukamoto, S. J. Greaves, K. Yoshida, and H. Muraoka, “Finite-element and micromagnetic modeling of write heads for shingled recording,” IEEE Transactions on Magnetics, vol. 46, no. 3, pp. 715–721, 2010.
[3] A. Aghayev, M. Shafaei, and P. Desnoyers, “Skylight—a window on shingled disk operation,” ACM Trans. Storage, vol. 11, oct 2015.
[4] M. Shafaei, M. H. Hajkazemi, P. Desnoyers, and A. Aghayev, “Modeling drive- managed smr performance,” ACM Trans. Storage, vol. 13, dec 2017.
[5] Y. Cassuto, M. A. A. Sanvido, C. Guyot, D. R. Hall, and Z. Z. Bandic, “Indirection systems for shingled-recording disk drives,” in 2010 IEEE 26th Symposium on Mass Storage Systems and Technologies (MSST), pp. 1–14, 2010.
[6] L. Ma and L. Xu, “Hmss: A high performance host-managed shingled storage sys- tem based on awareness of smr on block layer,” in 2016 IEEE 18th International Conference on High Performance Computing and Communications; IEEE 14th In-ternational Conference on Smart City; IEEE 2nd International Conference on Data Science and Systems (HPCC/SmartCity/DSS), pp. 570–577, 2016.
[7] F. Wu, M.-C. Yang, Z. Fan, B. Zhang, X. Ge, and D. H. Du, “Evaluating host aware SMR drives,” in 8th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage 16), (Denver, CO), USENIX Association, June 2016.
[8] T. Yao, Z. Tan, J. Wan, P. Huang, Y. Zhang, C. Xie, and X. He, “Sealdb: An ef- ficient lsm-tree based kv store on smr drives with sets and dynamic bands,” IEEE Transactions on Parallel and Distributed Systems, vol. 30, no. 11, pp. 2595–2607, 2019.
[9] S.-H. Chen, Y. Liang, and M.-C. Yang, “Kvstl: An application support to lsm-tree based key-value store via shingled translation layer data management,” IEEE Trans- actions on Computers, vol. 71, no. 7, pp. 1598–1611, 2022.
[10] M.-C. Yang, Y.-H. Chang, F. Wu, T.-W. Kuo, and D. H. Du, “Virtual persistent cache: Remedy the long latency behavior of host-aware shingled magnetic recording drives,” in 2017 IEEE/ACM International Conference on Computer-Aided Design (ICCAD), pp. 17–24, 2017.
[11] P. Xu, J. Wan, P. Huang, B. Shu, C. Tang, and C. Xie, “An active method to miti-gate the long latencies for host-aware shingle magnetic recording drives,” in 2019 IEEE 25th International Conference on Parallel and Distributed Systems (ICPADS), pp. 17–26, 2019.
[12] B. Zhang, M.-H. Yang, X. Xie, and D. H. Du, “Idler : I/o workload controlling for better responsiveness on host-aware shingled magnetic recording drives,” IEEE Transactions on Computers, vol. 69, no. 6, pp. 777–788, 2020.
[13] W. He and D. H. Du, “SMaRT: An approach to shingled magnetic recording trans- lation,” in 15th USENIX Conference on File and Storage Technologies (FAST 17), (Santa Clara, CA), pp. 121–134, USENIX Association, Feb. 2017.
[14] M. Shafaei and P. Desnoyers, “Virtual guard: A Track-Based translation layer for shingled disks,” in 9th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage 17), (Santa Clara, CA), USENIX Association, July 2017.
[15] Y.-J. Chuang, S.-H. Chen, Y.-H. Chang, Y.-P. Liang, H.-W. Wei, and W.-K. Shih, “Dstl: A demand-based shingled translation layer for enabling adaptive address map- ping on smr drives,” ACM Trans. Embed. Comput. Syst., vol. 19, jul 2020.
[16] M. H. Hajkazemi, M. Abdi, and P. Desnoyers, “ucache: a mutable cache for smr translation layer,” in 2020 28th International Symposium on Modeling, Analysis,and Simulation of Computer and Telecommunication Systems (MASCOTS), pp. 1–8, 2020.
[17] C. Ma, Y. Wang, Z. Shen, and Z. Shao, “Kfr: Optimal cache management with k- framed reclamation for drive-managed smr disks,” in 2020 57th ACM/IEEE Design Automation Conference (DAC), pp. 1–6, 2020.
[18] Y. Pan, Z. Jia, Z. Shen, B. Li, W. Chang, and Z. Shao, “Reinforcement learning- assisted cache cleaning to mitigate long-tail latency in dm-smr,” in 2021 58th ACM/ IEEE Design Automation Conference (DAC), pp. 103–108, 2021.
[19] D. Narayanan, A. Donnelly, and A. Rowstron, “Write off-loading: Practical power management for enterprise storage,” ACM Transactions on Storage (TOS), vol. 4, no. 3, pp. 1–23, 2008.
[20] K.-Y. CHEN, “Short-term and long-term idle time detectors for reducing long-tail latency in solid-state drives,” 2021.
[21] S.-J. Li, “A martingale and anti-martingale platform for idle time detection in solid- state drives,” 2022.