請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/92109
完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.advisor | 郭大維 | zh_TW |
dc.contributor.advisor | Tei-Wei Kuo | en |
dc.contributor.author | 艾俊賢 | zh_TW |
dc.contributor.author | Hasan Mohd Mamdouh Suleiman Alhasan | en |
dc.date.accessioned | 2024-03-05T16:19:50Z | - |
dc.date.available | 2024-07-05 | - |
dc.date.copyright | 2024-03-05 | - |
dc.date.issued | 2024 | - |
dc.date.submitted | 2024-02-16 | - |
dc.identifier.citation | [1] Samsung Electronics Co., Ltd. Samsung SSD 845DC Quality of Service (QoS) Guide. Technical report, Samsung Electronics, 2014. Accessed: 2024-02-05.
[2] Mingzhe Hao, Gokul Soundararajan, Deepavali Kenchammana-Hosekote, Andrew A Chien, and Haryadi S Gunawi. The tail at store: A revelation from millions of hours of disk and SSD deployments. In 14th USENIX Conference on File and Storage Technologies (FAST 16), pages 263–276, 2016. [3] Jieun Kim, Dongkun Lee, and Sam H Noh. Towards SLO complying SSDs through OPS isolation. In 13th USENIX Conference on File and Storage Technologies (FAST 15), pages 183–189, 2015. [4] HighPoint Technologies, Inc. Exploring the powerhouse: A deep dive into PCIe lane values of PCIe M.2 NVMe cards, 2020. Accessed: 2024-01-17. [5] Shuyi Pei, Jing Yang, and Qing Yang. Registor: A platform for unstructured data processing inside SSD storage. ACM Trans. Storage, 15(1), March 2019. [6] Arm. The guide to computational storage at Arm. White paper, Arm, 2020. [7] Hasan Alhasan, Yun-Chih Chen, and Chien-Chung Ho. RVO: Unleashing SSD's parallelism by harnessing the unused power. In 2021 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED), pages 1–6, 2021. [8] L. Chang, C. Cheng, S. Chang, and P. Chou. Current-aware flash scheduling for current capping in solid state disks. IEEE TCAD, 39(2):321–334, 2020. [9] Seongwook Jin, Jae-Hong Kim, Jaegeuk Kim, Jaehyuk Huh, and Seungryoul Maeng. Sector log: fine-grained storage management for solid state drives. In Proceedings of the ACM Symposium on Applied Computing, pages 360–367, 2011. [10] Mincheol Kang, Wonyoung Lee, and Soontae Kim. Subpage-based flash translation layer for solid state drivers. In Proceedings of the 2011 ACM Symposium on Applied Comp, 2016. [11] Synology Inc. Synology SSD Cache White Paper, 2022. Accessed: 2024-02-05. [12] Samsung Electronics Co., Ltd. Samsung SSD for data centers. Technical report, Samsung Electronics America, Inc., January 2015. Accessed: 2024-02-05. [13] Hasan Alhasan, Yun-Chih Chen, Chien-Chung Ho, and Tei-Wei Kuo. RUSM: Harnessing unused resources in 3D NAND SSD to enhance reading performance. In 2022 IEEE 11th Non-Volatile Memory Systems and Applications Symposium (NVMSA), pages 63–68, 2022. [14] Jung-Hoon Kim, Sang-Hoon Kim, and Jin-Soo Kim. Subpage programming for extending the lifetime of NAND flash memory. In 2015 Design, Automation & Test in Europe Conference & Exhibition (DATE), pages 555–560. IEEE, 2015. [15] Intel SSD DC S3700 series: Quality of service technical brief. https://tiscom.ru/sites/default/files/additional/ssd-dc-s3700-quality-service-tech-brief.pdf, 2013. Accessed: 2024-01-09. [16] Kingston DC400 SSD. https://media.kingston.com/pdfs/QOS-Quality-of-ServiceDC400-MKF-742.pdf, 2016. Accessed: 2024-02-05. [17] Suzhen Wu, Weiwei Zhang, Bo Mao, and Hong Jiang. Hotr: Alleviating read/write interference with hot read data replication for flash storage. In 2019 Design, Automation Test in Europe Conference Exhibition (DATE), pages 1367–1372, 2019. [18] Thomas Gibson. Understanding PCIe SSDs. https://www.pitsdatarecovery.com/what-is-pcie-ssd/, 2023. Accessed: 2024-01-04. [19] Kam Eshghi and Rino Micheloni. SSD Architecture and PCI Express Interface, pages 1–27. Springer Singapore, Singapore, 2018. [20] Seagate Technology LLC. Understanding the PCIe interface and how it benefits solid state storage. https://blog.seagate.com/intelligent/understanding-pcie-interface-benefits-solid-state-storage/, 2021. Accessed: 2024-01-04. [21] Donghyun Gouk, Jie Zhang, and Myoungsoo Jung. Enabling realistic logical device interface and driver for NVMe enabled full system simulations. International Journal of Parallel Programming, 46, August 2018. [22] Conner Carey. What is video metadata and how do I use it? https://sproutvideo.com/blog/what-is-video-metadata-and-how-do-i-use-it.html, 2023. Accessed: 2024-01-04. [23] Joe Danielson. Describing the details that matter: the importance of video metadata. https://www.axis.com/blog/secure-insights/video-metadata/, 2021. Accessed: 2024-01-04. [24] Gyusun Lee, Seokha Shin, Wonsuk Song, Tae Jun Ham, Jae W. Lee, and Jinkyu Jeong. Asynchronous I/O stack: A low-latency kernel I/O stack for Ultra-Low latency SSDs. In 2019 USENIX Annual Technical Conference (USENIX ATC 19), pages 603–616, Renton, WA, July 2019. USENIX Association. [25] Inode. https://en.wikipedia.org/wiki/Inode. Accessed: 2024-01-06. [26] Dave Mckay. Everything you ever wanted to know about inodes on Linux. https://www.howtogeek.com/465350/everything-you-ever-wanted-to-know-about-inodes-on-linux/, 2020. Accessed: 2024-01-04. [27] David Trounce. What are inodes in Linux and how are they used? https://helpdeskgeek.com/linux-tips/what-are-inodes-in-linux-and-how-are-they-used/, 2020. Accessed: 2024-01-06. [28] Dina Fakhry, Mohamed Abdelsalam, M. Watheq El-Kharashi, and Mona Safar. A review on computational storage devices and near memory computing for high performance applications. Memories- Materials, Devices, Circuits and Systems, 4:100051, 2023. [29] NVM Express, Inc. NVM Express Base Specification. https://nvmexpress.org/wp-content/uploads/NVMe-NVM-Express-2.0a-2021.07.26-Ratified.pdf, 2021. Accessed: 2024-01-06. [30] Aayush Gupta, Kim Youngjae, and Bhuvan Urgaonkar. DFTL: A flash translation layer employing demand-based selective caching of page-level address mappings. ACM Transactions on Storage, Volume 44, pages 229–240, February 2009. [31] Micron Technology, Inc. ECC in SLC NAND. Technical Note TN-29-63, Micron Technology, Inc., 2011. Accessed: 2024-01-09. [32] Matt Mills. How ECC error correction works on an SSD. https://www.itigic.com/how-ecc-error-correction-works-on-ssd/, 2020. Accessed: 2024-01-09. [33] See Jackson Hole. Town square live webcam. https://www.seejh.com/, Accessed 2024. Live surveillance video recording of street traffic. [34] OpenCV Team. Open source computer vision library. https://opencv.org/, Accessed 2024. Software library for computer vision and machine learning. [35] Joseph Redmon, Santosh Divvala, Ross Girshick, and Ali Farhadi. You only look once: Unified, real-time object detection. Proceedings of the IEEE conference on computer vision and pattern recognition, pages 779–788, 2016. [36] ONFI Working Group. Open NAND Flash Interface (ONFI) 3.0. ONFI Specification, 2011. | - |
dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/92109 | - |
dc.description.abstract | 這篇論文提出了優化固態硬盤(SSD)操作的新策略,重點關注三個主要領域:電源感知閃存命令重疊、主動熱數據複製和數據存儲與查詢的創新方法。這些技術旨在通過更好地利用SSD的資源來提升SSD的性能和效率,解決現代數據存儲系統中的挑戰。通過理論分析和實踐實施的結合,這項研究對SSD技術領域做出了重大貢獻,提供了改善數據處理和檢索的創新解決方案。 | zh_TW |
dc.description.abstract | This thesis presents novel strategies for optimizing Solid-State Drive (SSD) operations, focusing on three key areas: Power Aware Flash Command Overlapping, Proactive Hot Data Replication, and Innovative Approach to Data Storage and Querying. These techniques are designed to enhance SSD performance and efficiency via better utilization to SSD''s resources, addressing the challenges in modern data storage systems. Through a combination of theoretical analysis and practical implementation, this research contributes significantly to the field of SSD technology, offering innovative solutions to improve data processing and retrieval. | en |
dc.description.provenance | Submitted by admin ntu (admin@lib.ntu.edu.tw) on 2024-03-05T16:19:50Z No. of bitstreams: 0 | en |
dc.description.provenance | Made available in DSpace on 2024-03-05T16:19:50Z (GMT). No. of bitstreams: 0 | en |
dc.description.tableofcontents | Acknowledgements i
摘要 iii Abstract iv Contents v List of Figures x List of Tables xii Denotation xiii Chapter 1 Introduction 1 1.1 Quality of Service (QoS) in Data Centers 1 1.2 Background on SSD Internal Parallelism 1 1.3 SSD Constraints, and QoS Implications 2 1.4 Aiming for Enhanced SSD Performance with QoS Focus 4 1.5 Objectives 5 1.6 Research Contributions 5 1.7 Structure of the Dissertation 5 Chapter 2 Power Aware Flash Command Overlapping 6 2.1 Concept Overview 6 2.2 Background and Motivation 6 2.2.1 Power Constraints as a Bottleneck for SSD Parallel Processing 6 2.2.2 Data Recording in SSDs 7 2.2.3 Power under-utilization 8 2.3 Read-Verify Overlap (RVO) 8 2.3.1 Overview 8 2.3.2 Challenges 9 2.3.3 The Design of RVO Method 10 2.4 Scheduler Enhanced for RVO Integration 11 2.5 Implementation and Experimentation 12 2.6 Experimental Results 13 2.7 Conclusion 15 Chapter 3 Proactive Hot Data Replication 17 3.1 Concept Overview 17 3.2 Background and Motivation 18 3.2.1 The SSD''s Architecture 18 3.2.2 Address Translation and Subpage Writing in SSD 18 3.2.3 Inefficiencies in Subpage Writing Processes 18 3.2.4 Research Motivation 19 3.3 Replicate Using Subpage Merging (RUSM) 21 3.3.1 Overview 21 3.3.2 Challenges 22 3.3.3 The Design of RUSM Method 22 3.3.3.1 Request Workflows 22 3.3.4 Read-Cache Management in RUSM 23 3.3.4.1 Dual-List Structure 25 3.3.4.2 Pointer-Based Data Management 25 3.3.4.3 Hash Function Addressing 25 3.3.4.4 Sorting Overhead Management 26 3.3.4.5 Novelty of RUSM''s Read-Cache Mechanism 26 3.3.4.6 Advantages of RUSM''s Cache Design Over Regular SSD Cache 26 3.3.4.7 Cooperative Data Management with Operating System Caches 28 3.3.4.8 Subpage Consolidation and Address Translation in RUSM 30 3.3.4.9 Subpage Merging Process 30 3.3.4.10 Address Mapping Table 31 3.3.4.11 Space Overhead Management 31 3.4 Experimental Evaluation of RUSM 32 3.4.1 Related Work 32 3.4.2 Read Latency 33 3.4.3 Write Amplification Factor (WAF) 33 3.5 Conclusion 35 Chapter 4 Innovative Approach to Data Storage and Querying 36 4.1 Concept Overview 36 4.2 Background and Motivation 37 4.2.1 The I/O Stack Architecture in Solid-State Drives 37 4.2.2 SSD Data Management 38 4.2.3 Operational Layers Collaboration 38 4.2.4 Innovations in I/O Stack 38 4.2.5 Motivation: Conventional Video Retrieval 39 4.3 Information Storage and Data Framework (ISDF) 40 4.3.1 Overview 40 4.3.2 Challenges in Information Storage 42 4.3.3 Challenges in Data Framework 43 4.4 The Design of ISDF Methodology 44 4.4.1 Information Storage Design 44 4.4.2 Data Framework Design 47 4.4.2.1 Operational Workflow and Interaction 48 4.4.3 Repurposing ECC Engines for Pattern Recognition 50 4.4.3.1 ECC Encoding Process 50 4.4.3.2 ECC Decoding Process 51 4.4.3.3 Enhancement for Pattern Recognition 52 4.4.3.4 The Pattern Recognition Mode 52 4.4.3.5 ECC Data Segment Fitting for Pattern Recognition 53 4.4.3.6 IntegrationwithISDFFramework 54 4.4.3.7 Flash Memory Controller and ECC Mode Switching 55 4.4.3.8 Rationale Against False Positives in ECC-Based Pattern Recognition Mode 56 4.4.3.9 Limitations: Complex Queries and the Sufficiency of ECC Based Pattern Recognition 57 4.4.4 Video Segment Retrieval Using Information Storage 58 4.4.4.1 Metadata-Video Data Block Correspondence 58 4.4.4.2 Offset Calculation for Frame Retrieval 58 4.4.4.3 EfficientVideoContentDelivery 59 4.5 Experiments 59 4.5.1 Data Collection 59 4.5.2 Dataset Decomposition 60 4.5.3 Comparable Methods 61 4.5.4 Key Performance Index 62 4.5.5 Outcomes 62 4.5.6 Analysis 64 4.6 Conclusion 65 References 66 Appendix A - Implementation Costs for (RVO) Approach 71 A.1 PowerConsumptionin Command Transmission 71 A.2 FirmwareDevelopment 73 Appendix B - Implementation Costs for RUSM Approach 75 B.1 Power Consumption in DRAM 75 B.2 Assumption of Capacitors in High-Performing SSDs 76 Appendix C - Implementation Costs for ISDF Approach 78 C.1 ECC Modefor Pattern Recognition 78 C.2 Firmware Changes 78 C.3 CPU Workload 79 C.4 Interface Compatibility 79 | - |
dc.language.iso | en | - |
dc.title | 以快閃記憶體的並行性和處理能力提升系統性能研究 | zh_TW |
dc.title | Unleashing Flash Storage’s Parallelism and Processing Capability to Improve System Performance | en |
dc.type | Thesis | - |
dc.date.schoolyear | 112-1 | - |
dc.description.degree | 博士 | - |
dc.contributor.coadvisor | 何建忠 | zh_TW |
dc.contributor.coadvisor | Chien-Cheng Ho | en |
dc.contributor.oralexamcommittee | 劉邦鋒 ;施吉昇;洪士灝;張原豪 | zh_TW |
dc.contributor.oralexamcommittee | Pangfeng Liu;Chi-Sheng Shih;Shih-Hao Hung;Yuan-Hao Chang | en |
dc.subject.keyword | SSD 並行性,電源感知調度器,子頁面寫入,熱讀數據複製,讀取長尾延遲,資訊儲存,儲存內處理, | zh_TW |
dc.subject.keyword | SSD Parallelism,Power Aware Scheduler,Subpage Writing,Hot Read Data Replication,Read Long-Tail Latency,Information Storage,In-Storage Processing, | en |
dc.relation.page | 79 | - |
dc.identifier.doi | 10.6342/NTU202400563 | - |
dc.rights.note | 同意授權(全球公開) | - |
dc.date.accepted | 2024-02-17 | - |
dc.contributor.author-college | 電機資訊學院 | - |
dc.contributor.author-dept | 資訊工程學系 | - |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-112-1.pdf | 5.29 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。