中大機構典藏-NCU Institutional Repository-提供博碩士論文、考古題、期刊論文、研究計畫等下載:Item 987654321/93282
English  |  正體中文  |  简体中文  |  全文笔数/总笔数 : 80990/80990 (100%)
造访人次 : 41246857      在线人数 : 632
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜寻范围 查询小技巧:
  • 您可在西文检索词汇前后加上"双引号",以获取较精准的检索结果
  • 若欲以作者姓名搜寻,建议至进阶搜寻限定作者字段,可获得较完整数据
  • 进阶搜寻


    jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/93282


    题名: Prophet’s Insight: Unleashing Deduplication System Performance in Multi-tier Storage Systems
    作者: 拾以兆;Shih, Yi-Chao
    贡献者: 資訊工程學系
    关键词: 快閃記憶體;多層式儲存系統;重複資料刪除技術;NAND Flash;Multi-tier Storage System;Data Deduplication
    日期: 2023-07-26
    上传时间: 2024-09-19 16:52:27 (UTC+8)
    出版者: 國立中央大學
    摘要: 隨著巨量資料的增加,多層式儲存系統 (Multi-tier Storage System) 如傳統硬碟(Hard Disk Drive, HDD) 與固態硬碟 (Solid State Drive, SSD) 的組合被廣泛利用來增加成本效益。即使 SSD 的高速的讀寫特性能提高整體系統的效能,但 SSD 仍需面對其有限壽命、耐用程度的問題。為了解決這些問題,重複資料刪除 (Data Deduplication) 技術被提出,並且其可以有效減少 SSD 的寫入量。然而,隨著 SSD存儲媒介 NAND Flash 的基本單位 page 變大,會造成去重率 (Deduplication Rate)下降。本文提出了一種在 HDD-SSD 結合的儲存系統上的實施多粒度去重方法(Multi-grained Deduplication)。我們將 SSD 僅用作 read cache,並利用更細的粒度(例如: HDD 的基本單位,sector) 來進行去重。透過對 page 間的內容檢測,對於相似的 page 實施細粒度的去重,進而提高去重率、去重效能以及 I/O 效能。依據實驗結果顯示,與傳統方法 CAFTL 比較,多粒度去重在去重率上提高最多 69.4%,去重時間減少最多 4.8 倍。;As the volume of data continues to increase, multi-tier storage systems such as HDD-SSD combination are commonly adopted to achieve cost-effectiveness. Although SSDs can greatly benefit the overall system with their high perfor mance, they still face challenges related to limited lifespan and durability. To address these problems, data deduplication has been proposed as an effective
    technique to reduce the amount of data written to SSDs. However, as the size of NAND flash pages continues to grow, the deduplication rate with fixed-size chunking decreases. In this paper, we propose a multi-grained deduplication
    approach on the HDD-SSD combination storage system. We utilize SSD as solely a read cache and leveraging finer-granularity (i.e., sector) in deduplica tion. Specifically, our proposed solution minimizes computational overhead
    and achieves efficient detection on similar page content which contributes to fine-grained deduplication. By doing so, we can improve the deduplication rate, deduplication performance, and I/O performance. Our experimental re sults demonstrate that our multi-grained deduplication improves up to 69.4% in deduplication rate and reduces computational latency by up to 4.8 times with the compared baseline.
    显示于类别:[資訊工程研究所] 博碩士論文

    文件中的档案:

    档案 描述 大小格式浏览次数
    index.html0KbHTML6检视/开启


    在NCUIR中所有的数据项都受到原著作权保护.

    社群 sharing

    ::: Copyright National Central University. | 國立中央大學圖書館版權所有 | 收藏本站 | 設為首頁 | 最佳瀏覽畫面: 1024*768 | 建站日期:8-24-2009 :::
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 隱私權政策聲明