Abstract
Shingled magnetic recording (SMR) disks satisfy the growing demand for storage capacity for big data applications with their high capacity and low cost. However, the most severe challenge for SMR disks is the precipitous degradation of I/O performance when subjected to non-sequential writes. Using Solid State Drives (SSDs) as external caches for SMR disks is a cost-effective method to improve the I/O performance of SMR disks. Nevertheless, the existing SMR-oriented cache replacement algorithm is ineffective in resolving the conflict between the write amplification and the cache hit rate, resulting in limited performance gains from SSDs to SMR disks. In this paper, we design a multidevice cooperative buffer (MCB) management strategy to boost the write performance of SSD-SMR storage systems. Firstly, MCB selectively directs write traffic into the SSD cache to reduce the frequency of cleaning activity. Then, MCB adaptively evicts victim blocks from the SSD cache based on the locality principle. Finally, MCB utilizes a novelty SMR disk internal persistent buffer management mechanism to further optimize the efficiency of the SSD cache cleaning. The experimental results show that MCB achieves 7.4\(\times\) and 1.5\(\times\) performance improvements for write-intensive traces with spatial locality and temporal locality, respectively, compared to the state-of-the-art strategies.
Similar content being viewed by others
Data availability
All of the material is owned by the authors and/or no permissions are required, and all experiments can be reimplemented.
Notes
The manufacturer packs a certain number of tracks on the disk surface into bands/zones, which prevents the propagation of write amplification caused by sectors updates.
A set of blocks in a cache/buffer that belongs to the same zone is treated as a cluster. When a cache/buffer replacement occurs, the victim cluster is written back to the native storage area.
The cleaning activity of the current cache/buffer causes the cleaning activity of the next-level buffer to be triggered, which may cause the user I/O requests to be blocked for a long time.
The timestamp is self-increasing as the number of I/O requests accumulates.
A hot read block is one where the last access to the block was a read request, and where the read frequency is greater than the write frequency.
References
Richter H, Dobin A, Heinonen O, Gao K, Veerdonk R, Lynch R, Xue J, Weller D, Asselin P, Erden M, Brockie R (2006) Recording on bit-patterned media at densities of 1Tb/in\(^2\) and beyond. IEEE Trans Magn 42(10):2255–2260
Kryder M, Gage E, McDaniel T, Challener W, Rottmayer R, Ju G, Hsia Y-T, Erden M (2008) Heat assisted magnetic recording. Proc IEEE 96(11):1810–1835
Zhu J-G, Zhu X, Tang Y (2008) Microwave assisted magnetic recording. IEEE Trans Magn 44(1):125–131
White RL, New RMH, Pease RFW (1997) Patterned media: a viable route to 50 Gbit/in2 and Up for magnetic recording? IEEE Trans Magn 33(1):990–995
Cassuto Y, Sanvido MAA, Guyot C, Hall DR, Bandic ZZ (2010) Indirection systems for shingled-recording disk drives. In: Proceedings of the 26th IEEE Symposium on Mass Storage Systems and Technologies. Incline Village, NV, United States, pp 1–14
Greaves S, Kanai Y, Muraoka H (2009) Shingled recording for 2–3 Tbit/in\(^2\). IEEE Trans Magn 45(10):3823–3829
Chen C, Chen S, Liang Y, Chen T, Hsu T, Wei H, Shih W (2021) Facilitating external sorting on SMR-based large-scale storage systems. Futur Gener Comput Syst 116:333–348
Amer A, Long DDE, Miller EL, Paris J, Schwarz SJT (2010) Design issues for a shingled write disk system. In: Proceedings of the 26th IEEE Symposium on Mass Storage Systems and Technologies. Incline Village, NV, USA, pp 1–12
Aghayev A, Shafaei M, Desnoyers P ( 2015) Skylight-a window on shingled disk operation. In: Proceedings of the 13th USENIX Conference on File and Storage Technologies. Santa Clara, CA, United States, pp 135–149
Wu F, Fan Z, Yang M-C, Zhang B, Ge X, Du DHC (2017) Performance evaluation of host aware shingled magnetic recording (HA-SMR) drives. IEEE Trans Comput 66(11):1932–1945
Xie X, Xiao L, Du DHC (2019) ZoneTier: a zone-based storage tiering and caching co-design to integrate SSDs with SMR drives. ACM Trans Storage 15(3):1–25
Ma C, Zhou Z, Wang Y, Wang Y, Mao R (2022) MU-RMW: minimizing unnecessary RMW operations in the embedded flash with SMR disk. In: Proceedings of Design, Automation & Test in Europe Conference & Exhibition. Antwerp, Belgium, pp 490–495
Wang C, Wang D, Chai Y, Sun D ( 2017) Larger cheaper but faster: SSD-SMR hybrid storage boosted by a new SMR-oriented cache framework. In: Proceedings of the 33th IEEE Symposium on Mass Storage Systems and Technologies. Santa Clara, CA, United States, pp 1–16
Xiao W, Dong H, Ma L, Liu Z, Zhang Q (2016) HS-BAS: a hybrid storage system based on band awareness of shingled write disk. In: Proceedings of IEEE 34th International Conference on Computer Design. Phoenix, AZ, United States, pp 64–71
Liu W, Zeng L, Feng D, Kent KB (2019) ROCO: using a solid state drive cache to improve the performance of a host-aware shingled magnetic recording drive. J Comput Sci Technol 34(1):61–76
Xie X, Yang T, Li Q, Wei D, Xiao L (2018) Duchy: achieving both SSD durability and controllable SMR cleaning overhead in hybrid storage systems. In: Proceedings of the 47th IEEE International Conference on Parallel Processing. Eugene, Oregon, United States, pp 1–9
Sun D, Chai Y (2020) SAC: A co-design cache algorithm for emerging SMR-based high-density disks. In: Proceedings of the 25th International Conference on Architectural Support for Programming Languages and Operating Systems. Lausanne, Switzerland, pp 1047–1061
Yang M-C, Chang Y, Wu F, Kuo DW, Du DH (2018) On improving the write responsiveness for host-aware SMR drives. IEEE Trans Comput 68(1):111–124
Wu F, Li B, Du DHC (2021) FluidSMR: adaptive management for hybrid SMR drives. ACM Trans Storage (TOS) 17(4):1–30
Liang Y, Chen S, Chang Y, Lin Y, Wei H, Shih W (2019) Mitigating write amplification issue of SMR drives via the design of sequential-write-constrained cache. J Syst Architect 99:101634
Ma C, Shen Z, Wang Y, Shao Z (2018) Alleviating hot data write back effect for shingled magnetic recording storage systems. IEEE Trans Comput Aided Des Integr Circuits Syst 38(12):2243–2254
Yang T, Wu H, Huang P, Zhang F (2017) A shingle-aware persistent cache management scheme for DM-SMR disks. In: 2017 IEEE International Conference on Computer Design (ICCD). pp 81–88
Western Digital (2016) dm-zoned. Retrieved from https://github.com/hgst/dm-zoned-tools
Seagate Ships Worlds First 8TB Hard Drives, August 2014. http://www.seagate.com/about/newsroom/
Balakrishnan S, Black R, Donnelly A, England P, Glass A, Harper D, et al., (2014) Pelican: a building block for exascale cold data storage. In: Proceedings of the 11th USENIX Symposium on Operating Systems Design and Implementation. Broomfield, CO, United States, pp 351–365
Dropbox (2018) Extending magic pocket innovation with the first petabyte scale SMR drive deployment. https://blogs.dropbox.com/tech/2018/06/extending-magic-pocket-innovation-with-the-first-petabyte-scale-smr-drive-deployment/
INCITS T10 Technical Committee. [n. d.]. Information technology-Zoned Block Commands (ZBC). Draft Standard T10/BSR INCITS 536, American National Standards Institute, Inc., December 2015
Zhang B, Yang M, Xie X, Du DHC (2020) Idler : I/O workload controlling for better responsiveness on host-aware shingled magnetic recording drives. IEEE Trans Comput 69(6):777–788
Patana-anake T, Martin V, Sandler N, Wu C, Gunawi HS (2016) Manylogs: improved CMR/SMR disk bandwidth and faster durability with scattered logs. In: Proceedings of the 32nd Symposium on Mass Storage Systems and Technologies. Santa Clara, CA, United States, pp 1–16
Yao T, Wan J, Huang P, Zhang Y, Liu Z, Xie C, et al. (2019) GearDB: a GC-free key-value store on HM-SMR drives with gear compaction. In: Proceedings of the 17th USENIX Conference on File and Storage Technologies. Boston, MA, United States, pp 159–171
Aghayev A, Tso T, Gibson G, Desnoyers P (2017) Evolving ext4 for shingled disks. In: Proceedings of the 15th USENIX Conference on File and Storage Technologies. Santa Clara, CA, United States, pp 105–120
Jin C, Xi W, Ching Z, Huo F, Lim C (2014) HiSMRfs: a high performance file system for shingled storage array. In: Proceedings of the 30th IEEE Symposium on Mass Storage Systems and Technologies. Santa Clara, CA, United States, pp 1–6
Zeng L, Zhang Z, Wang Y, Feng D, Kent KB (2017) CosaFS: a cooperative shingle-aware file system. ACM Trans Storage 13(4):1–23
Zhang Y, Yao T, Wan J, Xie C (2022) Building GC-free key-value store on HM-SMR drives with ZoneFS. ACM Trans Storage (TOS) 18(3):1–23
Wu C, Yang M, Chang Y (2018) Improving runtime performance of deduplication system with host-managed SMR storage drives. In: 2018 55th ACM/ESDA/IEEE Design Automation Conference (DAC). pp 1–6
The Linux kernel user’s and administrator’s guide. https://docs.kernel.org/admin-guide/device-mapper/dm-zoned.html
Guerra J, Pucha H, Glider JS, Belluomini W, Rangaswami R (2015) Cost effective storage using extent based dynamic tiering. In: Proceedings of the 11th USENIX Conference on File and Storage Technologies. San Jose, CA, United States, pp 20–20
Luo D, Wan J, Zhu Y, Zhao N, Li F, Xie C (2016) Design and implementation of a hybrid shingled write disk system. IEEE Trans Parallel Distrib Syst 27(4):1017–1029
SDC 2020. https://www.snia.org/educational-library/high-performance-smr-drives-dm-zoned-caching-2020
Haghdoost A, He W, Fredin J, Du DH (2017) On the accuracy and scalability of intensive I/O workload replay. In: Proceedings of the 15th USENIX Conference on File and Storage Technologies. Santa Clara, CA, United States, pp 315-328
HGST (2017) Libzbc version 5.4.1. https://github.com/westerndigitalcorporation/libzbc
Narayanan D, Donnelly A, Rowstron A (2008) Write offloading: practical power management for enterprise storage. ACM Trans Storage 4(10):1–23
SNIA IOTTA Repository: MSR Cambridge Block I/O Traces, http://iotta.cs.hmc.edu/traces/388
Acknowledgements
This work was supported in part by the National Natural Science Foundation of China (61972311, 62002279) and in part by the Shandong Provincial Natural Science Foundation under Grant (ZR2021LZH009).
Funding
This work was supported in part by the National Natural Science Foundation of China (61972311, 62002279), in part by the Shandong Provincial Natural Science Foundation under Grant (ZR2021LZH009).
Author information
Authors and Affiliations
Contributions
CZ: Conceptualization, Methodology, Software, Investigation, Writing—original draft, Writing—review and editing, Validation. SN: Investigation. JW: Formal analysis. SL: Data curation. WW: Resources, Funding acquisition, Project administration, Supervision.
Corresponding author
Ethics declarations
Conflict of interest
I declare that the authors have no conflict of interest as defined by Springer, or other interests that might be perceived to influence the results and/or discussion reported in this paper.
Ethical approval
This paper does not contain any studies with human or animal subjects performed by any of the authors.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Zhang, C., Nie, S., Wang, J. et al. MCB: a multidevice cooperative buffer management strategy for boosting the write performance of the SSD-SMR hybrid storage. J Supercomput 79, 13462–13489 (2023). https://doi.org/10.1007/s11227-023-05201-7
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11227-023-05201-7