ABSTRACT
The performance of flash-based solid state drives is greatly impacted by the garbage collection algorithm. The d-choices garbage collection algorithm, which selects a victim block with the fewest number of valid pages among d randomly selected blocks, is known to perform well in terms of the write amplification. However, the number of erasures performed on a block may be quite unbalanced, which reduces the lifespan of SSDs that can only tolerate a limited number of erasures per block. This unequal wear is caused by the hot/cold data separation used to achieve a low write amplification, as blocks holding hot data tend to endure more erasures.
Methods to reduce this unequal wear often cause a significant increase in the write amplification (which slows down the device). In this paper we propose a new mechanism that allows us to severely reduce the unequal wear and thus improve the lifespan of the drive, without a significant increase in the write amplification. In fact, in many cases this mechanism even reduces the write amplification (eliminating the trade-off between low write amplification and a large lifespan altogether).
To assess the performance of this new mechanism we rely both on a mean field model and simulation experiments.
- M. Benaïm and J. Le Boudec. A class of mean field interaction models for computer and communication systems. Performance Evaluation, 65(11-12):823--838, 2008.Google ScholarDigital Library
- S. Boboila and P. Desnoyers. Write endurance in flash drives: Measurements and analysis. In Proceedings of the 8th USENIX Conference on File and Storage Technologies, FAST'10, pages 9-9, Berkeley, CA, USA, 2010. USENIX Association.Google ScholarDigital Library
- W. Bux and I. Iliadis. Performance of greedy garbage collection in flash-based solid-state drives. Perform. Eval., 67(11):1172--1186, November 2010.Google ScholarDigital Library
- Li-Pin Chang. On efficient wear leveling for large-scale flash-memory storage systems. In Proceedings of the 2007 ACM Symposium on Applied Computing, SAC '07, pages 1126-1130, New York, NY, USA, 2007. ACM.Google ScholarDigital Library
- Li-Pin Chang and Tei-Wei Kuo. An adaptive striping architecture for flash memory storage systems of embedded systems. In Proceedings of the Eighth IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS'02), RTAS 02, pages 187-, Washington, DC, USA, 2002. IEEE Computer Society.Google Scholar
- Mei-Ling Chiang, Paul C. H. Lee, and Ruei-Chuan Chang. Using data clustering to improve cleaning performance for flash memory. Softw. Pract. Exper., 29(3):267--290, March 1999.Google ScholarDigital Library
- P. Desnoyers. Analytic models of SSD write performance. ACM Trans. Storage, 10(2):8:1--8:25, March 2014.Google ScholarDigital Library
- E. Gal and S. Toledo. Algorithms and data structures for flash memories. ACM Computing Surveys, 37:138--163, 2005.Google ScholarDigital Library
- L. M. Grupp, J. D. Davis, and S. Swanson. The bleak future of NAND flash memory. In Proc. of USENIX Conference on File and Storage Technologies, 2012.Google ScholarDigital Library
- J. Hsieh, T. Kuo, and L. Chang. Efficient identification of hot data for flash memory storage systems. ACM Trans. on Storage, 2:22--40, 2006.Google ScholarDigital Library
- X. Hu, E. Eleftheriou, R. Haas, I. Iliadis, and R. Pletka. Write amplification analysis in flash-based solid state drives. In Proceedings of SYSTOR 2009: The Israeli Experimental Systems Conference, SYSTOR '09, pages 10:1--10:9, New York, NY, USA, 2009.Google ScholarDigital Library
- Yixin Luo, Yu Cai, Saugata Ghose, Jongmoo Choi, and Onur Mutlu. Warm: Improving nand flash memory lifetime with write-hotness aware retention management. In MSST, pages 1-14. IEEE Computer Society, 2015.Google ScholarCross Ref
- J. Menon. A performance comparison of RAID-5 and log-structured arrays. In Proceedings of the 4th IEEE International Symposium on High Performance Distributed Computing, HPDC '95, pages 167-178, Washington, DC, USA, 1995.Google ScholarDigital Library
- M. Murugan and D. Du. Rejuvenator: A static wear leveling algorithm for NAND flash memory with minimized overhead. In Proc. of IEEE MSST, 2011.Google ScholarDigital Library
- S. Odeh and Y. Cassuto. NAND flash architectures reducing write amplification through multi-write codes. In IEEE 30th Symposium on Mass Storage Systems and Technologies, MSST 2014, Santa Clara, CA, USA, June 2-6, 2014, pages 1-10, 2014.Google ScholarCross Ref
- D. Park and D. Du. Poster: Hot data identification for flash memory using multiple bloom filters. In Proc. of USENIX Conference on File and Storage Technologies, 2011.Google Scholar
- J.T. Robinson. Analysis of steady-state segment storage utilizations in a log-structured file system with least-utilized segment cleaning. SIGOPS Oper. Syst. Rev., 30(4):29--32, October 1996.Google ScholarDigital Library
- M. Rosenblum and J. K. Ousterhout. The design and implementation of a log-structured file system. ACM Trans. Comput. Syst., 10(1):26--52, February 1992.Google ScholarDigital Library
- 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., 41(1):191--202, 2013.Google ScholarDigital Library
- B. Van Houdt. Performance of garbage collection algorithms for flash-based solid state drives with hot/cold data. Perform. Eval., 70(10):692--703, 2013.Google ScholarDigital Library
- B. Van Houdt. On the necessity of hot and cold data identification to reduce the write amplification in flash-based SSDs. Perform. Eval., 82:1--14, 2014.Google ScholarDigital Library
- R. Verschoren and B. Van Houdt. On the endurance of the d-choices garbage collection algorithm for flash-based ssds. ACM Trans. Model. Perform. Eval. Comput. Syst., 4(3):13:1--13:23, July 2019.Google ScholarDigital Library
- Y. Yang, V. Misra, and D. Rubenstein. On the optimality of greedy garbage collection for SSDs. ACM SIGMETRICS Perform. Eval. Rev., 43(2):63--65, September 2015.Google ScholarDigital Library
Index Terms
- How to improve the performance of the d-choices garbage collection algorithm in flash-based SSDs
Recommendations
On the Endurance of the d-Choices Garbage Collection Algorithm for Flash-Based SSDs
Garbage collection (GC) algorithms for flash-based solid-state drives (SSDs) have a profound impact on its performance and many studies have focused on assessing the so-called write amplification of various GC algorithms. In this article, we consider ...
Age-based garbage collection
Modern generational garbage collectors look for garbage among the young objects, because they have high mortality; however, these objects include the very youngest objects, which clearly are still live. We introduce new garbage collection algorithms, ...
Analysis of the d-choices garbage collection algorithm with memory in flash-based SSDs
ValueTools '13: Proceedings of the 7th International Conference on Performance Evaluation Methodologies and ToolsGarbage collection algorithms have a profound impact on the performance and life span of flash-based solid state drives. Recently, the d-choices garbage collection algorithm was shown to provide an excellent tradeoff between simplicity and performance [...
Comments