Ada-WL: An Adaptive Wear-Leveling Aware Data Migration Approach for Flexible SSD Array Scaling in Clusters

Yunfei Gu; Linhui Liu; Chentao Wu; Jie Li; Minyi Guo

doi:10.1109/TC.2024.3398493

IEEE Transactions on Computers

Ada-WL: An Adaptive Wear-Leveling Aware Data Migration Approach for Flexible SSD Array Scaling in Clusters

Aug. 2024, pp. 1967-1982, vol. 73

DOI Bookmark: 10.1109/TC.2024.3398493

Authors

Yunfei Gu, Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
Linhui Liu, Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
Chentao Wu, Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
Jie Li, Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
Minyi Guo, Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China

Keywords

Arrays, Adaptive Arrays, Time Factors, Real Time Systems, Data Centers, Adaptation Models, Solid State Drives, Storage Scaling, Solid State Drive, Interdisk Wear Leveling, Data Migration, Object Object, Object Object Object Object, Object Object, Response Time, Data Center, Kalman Filter, Adaptive Control, Scaling Method, Adaptive Law, Average Response Time, Data Distribution, Simulation Experiments, Multiple Scales, Object Object Object Object, Storage Systems, Migration Process, Reference Model, Space Complexity Object Object, Object Object, Dynamic Adjustment, Temporal Localization Object Object, Hadoop Distributed File System, Kalman Gain, Mapping Table, Reduction In Response Time, Migration Data, Round Robin, Future Distribution, Fluctuations In Performance

Abstract

Recently, the flash-based Solid State Drive (SSD) array has been widely implemented in real-world large-scale clusters. With the increasing number of users in upper-tier applications and the burst of Input/Output requests in this data explosive era, data centers need to continuously scale up to meet real-time data storage needs. However, the classical disk array scaling methods are designed based on HDDs, ignoring the wear leveling and garbage collection characteristics of SSD. This leads to penalties due to the vast lifetime gap between extended SSDs and the original in-use SSDs while scaling the SSD array, including extra triggered wear leveling I/O, latency in average response time, etc. To address these problems, we propose an Adaptive Wear-Leveling aware data migration approach for flexible SSD array scaling in clusters. It manages the interdisk wear leveling based on Model Reference Adaptive Control, which includes an SSD behavior emulator, Kalman filter estimator, and adaptive law. To demonstrate the effectiveness of this approach, we conducted several simulations and implementations on actual hardware. The evaluation results show that Ada-WL has the self-adaptability to optimize the wear leveling management parameters for various states of SSD arrays, diverse workloads, and scaling performed multiple times, significantly improving performance for SSD array scaling.

Like what you’re reading?

Already a member?

Get this article FREE with a new membership!

References

[1]Y. Zhang , “Multi-view feature-based SSD failure prediction: What, when, and why,” in Proc. 21st USENIX Conf. File Storage Technol.(USENIX FAST’23), 2023, pp. 409–424.

[2]B. Schroeder, R. Lagisetty, and A. Merchant, “Flash reliability in production: The expected and the unexpected,” in Proc. 14th USENIX Conf. File Storage Technol. (FAST), vol. 16, 2016, pp. 67–80.

[3]“Microsoft Azure premium storage.” Microsoft Azure. Accessed: Apr., 2015. [Online]. Available: https://azure.microsoft.com/en-us/blog/azure-premium-storage-now-generally-available-2/

[4]M. Kang, D.-I. Kang, J. P. Walters, and S. P. Crago, “A comparison of system performance on a private OpenStack cloud and Amazon EC2,” in Proc. IEEE 10th Int. Conf. Cloud Comput. (CLOUD), 2017, pp. 310–317.

[5]Q. Li , “More than capacity: performance-oriented evolution of Pangu in Alibaba,” in Proc. 21st USENIX Conf. File Storage Technol. (FAST), vol. 23, 2023, pp. 331–346.

[6]S. Wang, Q. Cao, Z. Lu, H. Jiang, J. Yao, and Y. Dong, “StRAID: Stripe-threaded architecture for parity-based RAIDs with ultra-fast SSDs,” in Proc. USENIX Annu. Tech. Conf. (USENIX ATC 22), Carlsbad, CA, USA, 2022, pp. 915–932.

[7]M. Zwolenski and L. Weatherill, “The digital universe: Rich data and the increasing value of the internet of things,” J. Telecommun. Digit. Economy, vol. 2, no. 3, pp. 47.1–47.9, 2014.

[8]J. Li, Q. Wang, P. P. Lee, and C. Shi, “An in-depth analysis of cloud block storage workloads in large-scale production,” in Proc. IEEE Int. Symp. Workload Characterization (IISWC), Piscataway, NJ, USA: IEEE Press, 2020, pp. 37–47.

[9]Z. Lin, H. Guo, C. Wu, J. Li, G. Xue, and M. Guo, “Rack-scaling: An efficient rack-based redistribution method to accelerate the scaling of cloud disk arrays,” in Proc. IEEE Int. Parallel Distrib. Process. Symp. (IPDPS), Piscataway, NJ, USA: IEEE Press, 2021, pp. 892–901.

[10]B. Wilder, Cloud Architecture Patterns: Using Microsoft Azure. Sebastopol, CA, USA: O’Reilly Media, Inc., Sep.2012.

[11]Y. Niu, B. Luo, F. Liu, J. Liu, and B. Li, “When hybrid cloud meets flash crowd: Towards cost-effective service provisioning,” in Proc. IEEE Conf. Comput. Commun. (INFOCOM), Piscataway, NJ, USA: IEEE Press, 2015, pp. 1044–1052.

[12]H. Guo , “Lazy-WL: A wear-aware load balanced data redistribution method for efficient SSD array scaling,” in Proc. IEEE Int. Conf. Cluster Comput. (CLUSTER), Piscataway, NJ, USA: IEEE Press, 2021, pp. 157–168.

[13]J. Zhong, “PID controller tuning: A short tutorial,” Mech. Eng., Purdue Univ., 2006, pp. 1–10.

[14]D. A. Patterson, G. Gibson, and R. H. Katz, “A case for redundant arrays of inexpensive disks (RAID),” in Proc. ACM SIGMOD Int. Conf. Manage. Data (SIGMOD’88), New York, NY, USA: ACM, 1988, pp. 109–116.

[15]N. Brown, “Online RAID-5 resizing.” drivers/md/raid5.c in the source code of Linux kernel 2.6.18. Accessed: Sep.2016. [Online]. Available: http://www.kernel.org/

[16]J. S. Plank, “The RAID-6 liberation codes,” in Proc. 6th USENIX Conf. File Storage Technol. (USENIX FAST’08), USA: USENIX Association, Feb.2008.

[17]C.-W. Park and Y.-Y. Han, “A practical parity scheme for tolerating triple disk failures in raid architectures,” in Proc. 6th Asian Comput. Sci. Conf. Adv. Comput. Sci. (ASIAN ’00), Berlin, Heidelberg: Springer-Verlag, 2000, pp. 58–68.

[18]S. K. Mishra, S. K. Vemulapalli, and P. Mohapatra, “Dual-crosshatch disk array: A highly reliable hybrid-RAID architecture,” in Proc. Int. Conf. Parallel Process., vol. I, P. Banerjee, Ed., Boca Raton, FL, USA: CRC Press, Aug.1995, pp. 146–149.

[19]G. Zhang, J. Shu, W. Xue, and W. Zheng, “SLAS: An efficient approach to scaling round-robin striped volumes,” ACM Trans. Storage (TOS), vol. 3, no. 1, pp. 3–es, 2007.

[20]W. Zheng and G. Zhang, “FastScale: Accelerate raid scaling by minimizing data migration,” in Proc. 9th USENIX Conf. File Storage Technol. (FAST’11), 2011, pp. 1–11.

[21]C. Wu and X. He, “GSR: A global stripe-based redistribution approach to accelerate RAID-5 scaling,” in Proc. 41st Int. Conf. Parallel Process., Piscataway, NJ, USA: IEEE Press, 2012, pp. 460–469.

[22]C. Wu, X. He, J. Han, H. Tan, and C. Xie, “SDM: A stripe-based data migration scheme to improve the scalability of RAID-6,” in Proc. IEEE Int. Conf. Cluster Comput., Piscataway, NJ, USA: IEEE Press, 2012, pp. 284–292.

[23]G. Zhang, G. Wu, Y. Lu, J. Wu, and W. Zheng, “Xscale: Online X-code RAID-6 scaling using lightweight data reorganization,” IEEE Trans. Parallel Distrib. Syst., vol. 27, no. 12, pp. 3687–3700, Dec.2016.

[24]Y. Jiang, C. Wu, J. Li, and M. Guo, “BDR: A balanced data redistribution scheme to accelerate the scaling process of XOR-based triple disk failure tolerant arrays,” in Proc. IEEE 34th Int. Conf. Comput. Des. (ICCD), Piscataway, NJ, USA: IEEE Press, 2016, pp. 72–79.

[25]A. Miranda and T. Cortes, “CRAID: Online RAID upgrades using dynamic hot data reorganization,” in Proc. 12th USENIX Conf. File Storage Technol. (FAST’14), 2014, pp. 133–146.

[26]W. Wang, T. Xie, and A. Sharma, “SWANS: An interdisk wear-leveling strategy for RAID-0 structured SSD arrays,” ACM Trans. Storage (TOS), vol. 12, no. 3, pp. 1–21, 2016.

[27]D. Yimo, L. Fang, C. Zhiguang, and M. Xin, “WeLe-RAID: A SSD-based RAID for system endurance and performance,” in Proc. IFIP Int. Conf. Netw. Parallel Comput., New York, NY, USA: Springer-Verlag, 2011, pp. 248–262.

[28]J. Ou, J. Shu, Y. Lu, L. Yi, and W. Wang, “EDM: An endurance-aware data migration scheme for load balancing in SSD storage clusters,” in Proc. IEEE 28th Int. Parallel Distrib. Process. Symp., Piscataway, NJ, USA: IEEE Press, 2014, pp. 787–796.

[29]N. Zhao , “Chameleon: An adaptive wear balancer for flash clusters,” in Proc. IEEE Int. Parallel Distrib. Process. Symp. (IPDPS), Piscataway, NJ, USA: IEEE Press, 2018, pp. 1163–1172.

[30]C. Lee, T. Kumano, T. Matsuki, H. Endo, N. Fukumoto, and M. Sugawara, “Understanding storage traffic characteristics on enterprise virtual desktop infrastructure,” in Proc. of the 10th ACM Int. Syst. Storage Conf., ser. SYSTOR ’17, New York, NY, USA: Association for Computing Machinery, 2017. Available: https://doi.org/10.1145/3078468.3078479

[31]J. Gu, C. Wu, and J. Li, “HOTIS: A hot data identification scheme to optimize garbage collection of SSDs,” in Proc. IEEE Int. Symp. Parallel Distrib. Process. Appl./IEEE Int. Conf. Ubiquitous Comput. Commun. (ISPA/IUCC), Piscataway, NJ, USA: IEEE Press, 2017, pp. 331–3317.

[32]S. Patro and K. K. Sahu, “Normalization: A preprocessing stage,” 2015, arXiv:1503.06462.

[33]F. A. Gers, J. Schmidhuber, and F. Cummins, “Learning to forget: Continual prediction with LSTM,” Neural Comput., vol. 12, no. 10, pp. 2451–2471, 2000.

[34]T. Kuremoto, M. Furuya, S. Mabu, and K. Kobayashi, “A time series forecasting method using DBN and adam optimization,” in Proc. Int. Conf. Artif. Intell. Commun. Netw., New York, NY, USA: Springer-Verlag, 2022, pp. 95–106.

[35]C. Pezzato, R. Ferrari, and C. H. Corbato, “A novel adaptive controller for robot manipulators based on active inference,” IEEE Robot. Automat. Lett., vol. 5, no. 2, pp. 2973–2980, Apr.2020.

[36]Y. Yang, V. Misra, and D. Rubenstein, “On the optimality of greedy garbage collection for SSDs,” ACM SIGMETRICS Perform. Eval. Rev., vol. 43, no. 2, pp. 63–65, 2015.

[37]Y. Pan, M. Lin, Z. Wu, H. Zhang, and Z. Xu, “Caching-aware garbage collection to improve performance and lifetime for NAND flash SSDs,” IEEE Trans. Consum. Electron., vol. 67, no. 2, pp. 141–148, May2021.

[38]R. Verschoren and B. V. Houdt, “On the endurance of the d-choices garbage collection algorithm for flash-based SSDs,” ACM Trans. Model. Perform. Eval. Comput. Syst. (TOMPECS), vol. 4, no. 3, pp. 1–23, 2019.

[39]B. Van Houdt, “A mean field model for a class of garbage collection algorithms in flash-based solid state drives,” ACM SIGMETRICS Perform. Eval. Rev., vol. 41, no. 1, pp. 191–202, 2013.

[40]G. Welch , “An introduction to the Kalman filter,” University of North Carolina at Chapel Hill, Department of computer science, TR, pp. 41–95, 1995.

[41]J.-W. Jung, V. Q. Leu, T. D. Do, E.-K. Kim, and H. H. Choi, “Adaptive PID speed control design for permanent magnet synchronous motor drives,” IEEE Trans. Power Electron., vol. 30, no. 2, pp. 900–908, Feb.2015.

[42]V. Prabhakaran and T. Wobber, “SSD Extension for DiskSim Simulation Environment,” Microsoft Research, 2008. Accessed: Nov., 2020. [Online]. Available: http://research.microsoft.com/en-us/downloads/b41019e2-1d2b-44d8-b512-ba35ab814cd4

[43]Apache Software Foundation, “Hadoop 3.3.4.” Accessed: Jul.2022. [Online]. Available: https://hadoop.apache.org/docs/r3.3.4/

[44]A. Tavakkol, J. Gómez-Luna, M. Sadrosadati, S. Ghose, and O. Mutlu, “MQSim: A framework for enabling realistic studies of modern multi-queue SSD devices,” in Proc. 16th USENIX Conf. File Storage Technol. (FAST), Feb.2018.

Making Garbage Collection Wear Conscious for Flash SSD
2012 IEEE Seventh International Conference on Networking, Architecture, and Storage
DCS5: Diagonal Coding Scheme for Enhancing the Endurance of SSD-Based RAID-5 Systems
2014 9th IEEE International Conference on Networking, Architecture, and Storage (NAS)
WL-Reviver: A Framework for Reviving any Wear-Leveling Techniques in the Face of Failures on Phase Change Memory
2014 44th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN)
Error rate-based wear-leveling for NAND flash memory at highly scaled technology nodes
IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Adaptive Wear-Leveling in Flash-Based Memory
IEEE Computer Architecture Letters
RETROFIT: Fault-Aware Wear Leveling
IEEE Computer Architecture Letters
Efficient Wear Leveling for PCM/DRAM-Based Hybrid Memory
2019 IEEE 21st International Conference on High Performance Computing and Communications; IEEE 17th International Conference on Smart City; IEEE 5th International Conference on Data Science and Systems (HPCC/SmartCity/DSS)
Pattern-based Write Scheduling and Read Balance-oriented Wear-Leveling for Solid State Drivers
2019 35th Symposium on Mass Storage Systems and Technologies (MSST)
Comparative Study of Wear-leveling in Solid-State Drive with NTFS File System
2019 IEEE International Conference on Big Data (Big Data)
Lazy-WL: A Wear-aware Load Balanced Data Redistribution Method for Efficient SSD Array Scaling
2021 IEEE International Conference on Cluster Computing (CLUSTER)

Ada-WL: An Adaptive Wear-Leveling Aware Data Migration Approach for Flexible SSD Array Scaling in Clusters

Authors

Keywords

Abstract

References

Related Articles