服务器内存可靠性技术研究综述

doi:10.3778/j.issn.1002-8331.2409-0342

摘要/Abstract

摘要： 内存作为服务器中的核心部件，随着内存技术的持续迭代与性能的显著提升，其可靠性问题已成为影响服务器整体稳定性不可忽视的关键因素。回顾了内存技术的演进历程、结构特性及其发展对服务器性能的直接影响，深入剖析了内存故障模式的多样性与深层次复杂性。进一步地，详尽探讨了故障检测与处理的最新技术进展，特别强调了内存纠错码和内存故障容错技术的重要作用，并聚焦于内存风险单元预测技术的前沿探索，尤其是结合确定性规则或机器学习算法的内存故障预测方法。在此基础上，对当前内存可靠性领域面临的核心挑战进行了系统性分析，并前瞻性地展望了未来研究方向，涵盖内存老化精准预测、健康状态实时监测及机器学习在预测分析中的深度应用等关键领域。最终强调，在追求服务器内存性能极致化的同时，必须并行提升其稳定性与可靠性，以适应日益增长的服务器性能需求，为内存可靠性技术的未来发展提供了宝贵的实践指导与理论参考。

关键词: 服务器内存, 故障容错, 故障检测, 健康监测, 机器学习

Abstract: Memory, as a core component in servers, has experienced continuous iterations in technology and significant improvements in performance, yet its reliability issues have emerged as a critical factor that cannot be overlooked in influencing the overall stability of servers. This paper reviews the evolution of memory technology, its structural characteristics, and the direct impact of its development on server performance. It delves into the diversity and deep complexity of memory failure modes. Furthermore, the article comprehensively explores the latest technological advancements in fault detection and handling, with particular emphasis on the pivotal role of error correction codes (ECC) and RAS (reliability, availability, serviceability) technologies for memory. It also focuses on the forefront exploration of memory risk cell prediction technologies, particularly memory failure prediction methods that integrate deterministic rules or machine learning algorithms. Based on this foundation, the paper conducts a systematic analysis of the core challenges facing the current memory reliability field and offers a forward-looking outlook on future research directions, encompassing precise prediction of memory aging, real-time monitoring of health status, and the profound application of machine learning in predictive analysis. Ultimately, the paper underscores the necessity of concurrently enhancing the stability and reliability of server memory while pursuing its ultimate performance, to accommodate the ever-growing demands on server performance. This provides invaluable practical guidance and theoretical references for the future development of memory RAS technologies.

Key words: server memory, fault tolerance, fault detection, health monitoring, machine learning

李道童, 李盛新, 王兵, 姚藩益, 芦飞, 艾山彬, 张炳会, 孙秀强, 王若琳. 服务器内存可靠性技术研究综述[J]. 计算机工程与应用, 2025, 61(15): 72-92.

LI Daotong, LI Shengxin, WANG Bing, YAO Fanyi, LU Fei, AI Shanbin, ZHANG Binghui, SUN Xiu-qiang, WANG Ruolin. Review of Server Memory Reliability Technology[J]. Computer Engineering and Applications, 2025, 61(15): 72-92.

参考文献

[1] BOGATINOVSKI J, KAO O, YU Q, et al. First CE matters: on the importance of long term properties on memory failure prediction[C]//Proceedings of the 2022 IEEE International Conference on Big Data. Piscataway: IEEE, 2022: 4733-4736.
[2] DAYANAND N, QUAH A C T, CHEN C Q, et al. Static fault localization on memory failures using Photon Emission Microscopy[C]//Proceedings of the 2015 IEEE 22nd International Symposium on the Physical and Failure Analysis of Integrated Circuits. Piscataway: IEEE, 2015: 322-326.
[3] 王群, 李馥娟, 倪雪莉, 等. 域间路由安全增强及区块链技术的应用研究[J]. 计算机科学与探索, 2024, 18(12): 3144-3174.
WANG Q, LI F J, NI X L, et al. Research on blockchain-based inter-domain routing security enhancement[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(12): 3144-3174.
[4] 张格毅, 陈小刚, 郭继鹏, 等. 利用相变存储器不对称性的写入优化方法[J]. 计算机工程与应用, 2021, 57(14): 75-82.
ZHANG G Y, CHEN X G, GUO J P, et al. Solution to optimize PCM write depending on asymmetries[J]. Computer Engineering and Applications, 2021, 57(14): 75-82.
[5] LEE J, KIM M J, KIM W S, et al. Review of memory RAS for data centers[J]. IEEE Access, 2023, 11: 124782-124796.
[6] KIM D, KIM J. Adaptive granularity on-die ECC[C]//Proceedings of the 2022 19th International SoC Design Conference. Piscataway: IEEE, 2022: 318-319.
[7] CHENG Z N, HAN S J, LEE P P C, et al. An in-depth correlative study between DRAM errors and server failures in production data centers[C]//Proceedings of the 2022 41st International Symposium on Reliable Distributed Systems. Piscataway: IEEE, 2022: 262-272.
[8] FEVGAS A, AKRITIDIS L, ALAMANIOTIS M, et al. HyR-tree: a spatial index for hybrid flash/3D XPoint storage[J]. Neural Computing and Applications, 2023, 35(1): 133-145.
[9] CARNIEL A C, AGUIAR C D. Spatial index structures for modern storage devices: a survey[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(9): 9578-9597.
[10] SMAGULOVA K, FOUDA M E, ELTAWIL A. Thermal heating in ReRAM crossbar arrays: challenges and solutions[J]. IEEE Open Journal of Circuits and Systems, 2024, 5: 28-41.
[11] LEE S, LEE N H, LEE K, et al. Development and product reliability characterization of advanced high speed 14nm DDR5 DRAM with on-die ECC[C]//Proceedings of the 2023 IEEE International Reliability Physics Symposium. Piscataway: IEEE, 2023: 1-4.
[12] SEO H, RIM T, LEE E, et al. Analysis of intermittent single-bit failure on 10-nm node generation DRAM devices[C]//Proceedings of the 2023 IEEE International Reliability Physics Symposium. Piscataway: IEEE, 2023: 1-6.
[13] SPESSOT A, OH H. 1T-1C dynamic random access memory status, challenges, and prospects[J]. IEEE Transactions on Electron Devices, 2020, 67(4): 1382-1393.
[14] BRACKMANN L, JAFARI A, BENGEL C, et al. A failure analysis framework of ReRAM in-memory logic operations[C]//Proceedings of the 2022 IEEE International Test Conference in Asia. Piscataway: IEEE, 2022: 67-72.
[15] SRIDHARAN V, DEBARDELEBEN N, BLANCHARD S, et al. Memory errors in modern systems: the good, the bad, and the ugly[C]//Proceedings of the Twentieth International Conference on Architectural Support for Programming Languages and Operating Systems. New York: ACM, 2015: 297-310.
[16] ZHANG P C, WANG Y N, MA X H, et al. Predicting DRAM-caused node unavailability in hyper-scale clouds[C]//Proceedings of the 2022 52nd Annual IEEE/IFIP International Conference on Dependable Systems and Networks. Piscataway: IEEE, 2022: 275-286.
[17] MAYUGA G, YAMATO Y, YONEDA T, et al. Reliability-enhanced ECC-based memory architecture using in-field self-repair[J]. IEICE Transactions on Information and Systems, 2016, 99(10): 2591-2599.
[18] 许健, 陈平华, 熊建斌. 基于抽象内存模型的内存相关漏洞检测方法[J]. 计算机工程与应用, 2022, 58(8): 96-108.
XU J, CHEN P H, XIONG J B. Memory-related vulnerability detection method based on abstract memory model[J]. Computer Engineering and Applications, 2022, 58(8): 96-108.
[19] 路杉杉. March算法研究及其分析软件实现[D]. 成都: 电子科技大学, 2023.
LU S S. March algorithm study and its analysis software implementation[D]. Chengdu: University of Electronic Science and Technology of China, 2023.
[20] 王恩笙. 一种March类存储器测试序列解析方法的设计与实现[D]. 成都: 电子科技大学, 2024.
WANG E S. Design and implementation of a method for analysis of march-type memory test sequences[D]. Chengdu: University of Electronic Science and Technology of China, 2024.
[21] 葛云侠, 陈龙, 解维坤, 等. 大规模芯片内嵌存储器的BIST测试方法研究[J]. 国外电子测量技术, 2024, 43(5): 18-25.
GE Y X, CHEN L, XIE W K, et al. Research on BIST testing method for large-scale chip embedded memory[J]. Foreign Electronic Measurement Technology, 2024, 43(5): 18-25.
[22] BLAKE I F. Error control coding (S. Lin and D. J. Costello; 2004) [book review][J]. IEEE Transactions on Information Theory, 2005, 51(4): 1616-1617.
[23] KWON J H, BAE H K, LEE Y S, et al. ZEC ECC: a zero-byte eliminating compression-based ECC scheme for DRAM reliability[J]. IEEE Access, 2024, 12: 100366-100376.
[24] BAE H K, CHUNG M J, GONG Y H, et al. Twin ECC: a data duplication based ECC for strong DRAM error resilience[C]//Proceedings of the 2023 Design, Automation & Test in Europe Conference & Exhibition. Piscataway: IEEE, 2023: 1-6.
[25] LIN S, COSTELLO D J. Error control coding[M]. USA: Prentice-Hall, 2004: 357-762.
[26] REVIRIEGO P, LIU S S, XIAO L Y, et al. An efficient single and double-adjacent error correcting parallel decoder for the (24, 12) extended golay code[J]. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2016, 24(4): 1603-1606.
[27] BENTOUTOU Y. A real time EDAC system for applications onboard earth observation small satellites[J]. IEEE Transactions on Aerospace and Electronic Systems, 2012, 48(1): 648-657.
[28] LEE H, YOO Y, SHIN S H, et al. ECMO: ecc architecture reusing content-addressable memories for obtaining high reliability in DRAM[J]. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2022, 30(6): 781-793.
[29] HAYAKAWA A, NAKAMURA T, DEGUCHI Y, et al. Data-aware partial ECC with data modulation of ReRAM with non-volatile in-memory computing for image recognition with deep neural network[C]//Proceedings of the 2018 IEEE International Symposium on Circuits and Systems. Piscataway: IEEE, 2018: 1-5.
[30] DAS H, AHMAD HAIDOUS A, SMITH S C, et al. Flexible low-cost power-efficient video memory with ECC-adaptation[J]. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2021, 29(10): 1693-1706.
[31] KIM K, LIM S H. Compression and variable-sized ECC scheme for the reliable flash memory system[C]//Advances in Computer Science and Ubiquitous Computing. Singapore: Springer Singapore, 2017: 1232-1236.
[32] ALACCHI A, GIACOMIN E, TEMPLE S, et al. Low latency SEU detection in FPGA CRAM with in-memory ECC checking[J]. IEEE Transactions on Circuits and Systems I: Regular Papers, 2023, 70(5): 2028-2036.
[33] WU S Z, DU C F, ZHU W D, et al. EaD: ecc-assisted deduplication with high performance and low memory overhead for ultra-low latency flash storage[J]. IEEE Transactions on Computers, 2023, 72(1): 208-221.
[34] 田欢. 低冗余存储器相邻双错误纠正码设计[D]. 哈尔滨: 哈尔滨工业大学, 2011.
TIAN H. Low redundancy adjacent double: errors correction codes in memory[D]. Harbin: Harbin Institute of Technology, 2011.
[35] 王丹宁. 面向连续多位翻转的纠错码研究与实现[D]. 长沙: 国防科技大学, 2021.
WANG D N. The research and implementation of error correction code for continuous multi-bit upsets[D]. Changsha: National University of Defense Technology, 2021.
[36] 安天乐. 高效片上存储纠错技术研究与实现[D]. 长沙: 国防科技大学, 2018.
AN T L. The research and implementation of efficient error correction technology for on-chip memory[D]. Changsha: National University of Defense Technology, 2018.
[37] LI J, REVIRIEGO P, XIAO L, et al. Protecting memories against soft errors: the case for customizable error correction codes[J]. IEEE Transactions on Emerging Topics in Computing, 2021, 9(2): 651-663.
[38] BENSO A, CHIUSANO S, DI NATALE G, et al. An on-line BIST RAM architecture with self-repair capabilities[J]. IEEE Transactions on Reliability, 2002, 51(1): 123-128.
[39] BUCH S. Error detecting and correcting codes for DRAM functional safety[C]//Proceedings of the International Symposium on Memory Systems. New York: ACM, 2023: 1-5.
[40] DU X M, LI C. Predicting uncorrectable memory errors from the correctable error history: no free predictors in the field[C]//Proceedings of the International Symposium on Memory Systems. New York: ACM, 2021: 1-10.
[41] DELL T J. System RAS implications of DRAM soft errors[J]. IBM Journal of Research and Development, 2008, 52(3): 307-314.
[42] KIM J, KWON S, NOH J, et al. Construction of cyclic redundancy check codes for SDDC decoding in DRAM systems[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2023, 70(2): 736-740.
[43] LI W, ZHANG M, GUI T W, et al. Improving DRAM reliability using a high order error correction code[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2024, 43(12): 4775-4785.
[44] WARNES L, CALHOUN M B, CARR D, et al. Rank sparing system and method: US8892942[P]. 2014-11-18.
[45] DU X M, LI C, ZHOU S, et al. Predicting uncorrectable memory errors for proactive replacement: an empirical study on large-scale field data[C]//Proceedings of the 2020 16th European Dependable Computing Conference. Piscataway: IEEE, 2020: 41-46.
[46] GIURGIU I, SZABO J, WIESMANN D, et al. Predicting DRAM reliability in the field with machine learning[C]//Proceedings of the 18th ACM/IFIP/USENIX Middleware Conference: Industrial Track. New York: ACM, 2017: 15-21.
[47] BOIXADERAS I, ZIVANOVIC D, MORé S, et al. Cost-aware prediction of uncorrected DRAM errors in the field[C]//Proceedings of the SC20: International Conference for High Performance Computing, Networking, Storage and Analysis. Piscataway: IEEE, 2020: 1-15.
[48] CRISS K, BAINS K, AGARWAL R, et al. Improving memory reliability by bounding DRAM faults: ddr5 improved reliability features[C]//Proceedings of the International Symposium on Memory Systems. New York: ACM, 2020: 317-322.
[49] MEZA J, WU Q, KUMAR S, et al. Revisiting memory errors in large-scale production data centers: analysis and modeling of new trends from the field[C]//Proceedings of the 2015 45th Annual IEEE/IFIP International Conference on Dependable Systems and Networks. Piscataway: IEEE, 2015: 415-426.
[50] DU X M, LI C, ZHOU S, et al. Fault-aware prediction-guided page offlining for uncorrectable memory error prevention[C]//Proceedings of the 2021 IEEE 39th International Conference on Computer Design. Piscataway: IEEE, 2021: 456-463.
[51] DU X M, LI C. DPCLS: improving partial cache line sparing with dynamics for memory error prevention[C]//Proceedings of the 2020 IEEE 38th International Conference on Computer Design. Piscataway: IEEE, 2020: 197-204.
[52] DU X M, LI C. Combining error statistics with failure prediction in memory page offlining[C]//Proceedings of the International Symposium on Memory Systems. New York: ACM, 2019: 127-132.
[53] COSTA C H A, PARK Y, ROSENBURG B S, et al. A system software approach to proactive memory-error avoidance[C]//Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis. Piscataway: IEEE, 2014: 707-718.
[54] LI C, ZHANG Y, WANG J L, et al. From correctable memory errors to uncorrectable memory errors: what error bits tell[C]//Proceedings of the SC22: International Conference for High Performance Computing, Networking, Storage and Analysis. Piscataway: IEEE, 2022: 1-14.
[55] JUNG J, EREZ M. Predicting future-system reliability with a component-level DRAM fault model[C]//Proceedings of the 2023 56th IEEE/ACM International Symposium on Microarchitecture. Piscataway: IEEE, 2023: 944-956.
[56] BREITENBACH T, MALAVALLI DIVAKAR S, RASBACH L, et al. ML-driven risk estimation for memory failure in a data center environment with convolutional neural networks, self-supervised data labeling and distribution-based model drift determination[J]. Journal of Parallel and Distributed Computing, 2024, 185: 104800.
[57] YU Q, ZHANG W G, NOTARO P, et al. HiMFP: hierarchical intelligent memory failure prediction for cloud service reliability[C]//Proceedings of the 2023 53rd Annual IEEE/IFIP International Conference on Dependable Systems and Networks. Piscataway: IEEE, 2023: 216-228.
[58] BASEMAN E, DEBARDELEBEN N, FERREIRA K, et al. Improving DRAM fault characterization through machine learning[C]//Proceedings of the 2016 46th Annual IEEE/IFIP International Conference on Dependable Systems and Networks Workshop. Piscataway: IEEE, 2016: 250-253.
[59] BASEMAN E, DEBARDELEBEN N, FERREIRA K, et al. Automating DRAM fault mitigation by learning from experience[C]//Proceedings of the 2017 47th Annual IEEE/IFIP International Conference on Dependable Systems and Networks Workshops. Piscataway: IEEE, 2017: 137-140.
[60] GIURGIU I, WIESMANN D, BIRD J. “Memory loss” in commodity hardware : predicting DIMM failures with machine learning[C]//Proceedings of the 10th ACM International Systems and Storage Conference. New York: ACM, 2017.
[61] YU Q, CARDOSO J, KAO O. Unveiling DRAM failures across different CPU architectures in large-scale datacenters[C]//Proceedings of the 2024 IEEE 44th International Conference on Distributed Computing Systems. Piscataway: IEEE, 2024: 1462-1463.
[62] YU F Y, XU H Z, JIAN S L, et al. DRAM failure prediction in large-scale data centers[C]//Proceedings of the 2021 IEEE International Conference on Joint Cloud Computing. Piscataway: IEEE, 2021: 1-8.
[63] WANG X Y, LI Y, CHEN Y Q, et al. On workload-aware DRAM failure prediction in large-scale data centers[C]//Proceedings of the 2021 IEEE 39th VLSI Test Symposium. Piscataway: IEEE, 2021: 1-6.
[64] DU X M, LI C. Memory failure prediction using online learning[C]//Proceedings of the International Symposium on Memory Systems. New York: ACM, 2018: 38-49.
[65] BASEMAN E, DEBARDELEBEN N, BLANCHARD S, et al. Physics-informed machine learning for DRAM error modeling[C]//Proceedings of the 2018 IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems. Piscataway: IEEE, 2018: 1-6.
[66] ZHENG X, WU L Z, DONG D N, et al. Endurance prediction based on hidden Markov model and programming optimization for 28 nm 1 Mbit resistive random access memory chip[J]. IEEE Electron Device Letters, 2023, 44(6): 919-922.
[67] 李盛新, 李道童, 贾帅帅, 等. 内存状态检测方法、装置、通信设备及存储介质: CN202310935420. 8[P]. 2023-11-03.
LI S X, LI D T, JIA S S, et al. Memory state detection method and apparatus, and communication device and storage medium: CN202310935420. 8[P]. 2023-11-03.
[68] WU R L, ZHOU S Y, LU J H, et al. Removing obstacles before breaking through the memory wall: a close look at HBM errors in the field[C]//Proceedings of the 2024 Usenix Annual Technical Conference, 2024: 851-867.
[69] WANG J, YANG H Z, LI C, et al. Boosting data center performance via intelligently managed multi-backend disaggregated memory[C]//Proceedings of the SC24: International Conference for High Performance Computing, Networking, Storage and Analysis. Piscataway: IEEE, 2024: 1-18.
[70] LI P F, HUA Y, ZUO P F, et al. A high-performance RDMA-oriented learned key-value store for disaggregated memory systems[J]. ACM Transactions on Storage, 2023, 19(4): 1-30.
[71] LI H C, BERGER D S, HSU L, et al. Pond: cxl-based memory pooling systems for cloud platforms[C]//Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2. New York: ACM, 2023: 574-587.
[72] LEE H, CHOI K, LEE H J, et al. SDM: sharing-enabled disaggregated memory system with cache coherent compute express link[C]//Proceedings of the 2023 32nd International Conference on Parallel Architectures and Compilation Techniques, 2023: 86-98.