科技的進步飛快，資料儲存的問題已由過去的「如何存放大量資料」演變成現今的「如何從大量資料中找到想要的資訊」，因此資料的管理課題日漸浮現，資訊生命週期管理(ILM，Information Lifecycle Management)成為儲存管理中的重要議題，ILM意指從抓取、創造、保留、備份、到銷毀資料的整個過程並依據資料存取頻繁度、新舊、或是重要性來將資料搬移到適當的媒體上，在ILM中，如何有效地搬移成為舉足輕重的關鍵，搬移策略的制定扮演著ILM的要角，多個互相獨立的檔案若依照搬移策略，可有效地進行移動至階層式儲存裝置的其他級別的儲存媒體，但倘若搬移的對象為同一檔案的多個版本資料，由於如快照之類的資料保護技術特性使得這些資料為非互相獨立的，在邏輯上，利用快照所製作出來的多個版本是可以存成多個檔案，但是在實體上這多個檔案卻有部分的資料是共用的，因此在搬移的時候便會產生轉移不當的問題，而造成日後讀取新版檔案時效能下降的問題，但若以複製共用部分的資料來克服此問題則會造成資料重複。
因此為了改善此一課題，本論文提出快照的生命週期管理(Snapshot Lifecycle Management；SLM)，針對快照特性的檔案進行更為適宜且有效的管理。並更進一步的提出在異地備援下的遠端快照，可改善網路延遲所帶來的傳輸效能下降，藉由修改CoW (Copy-on-Write)機制以適用於廣域網路下的資料備份。SLM的適用範圍並不只限於快照，更可拓展到所有屬於增量備份型的備份機制，用來管理中程的備份儲存系統。

With the great and rapid advancement in Information Technology, the main issue in storage area turns from "how to store great quantity data" into "how to find the desired information from great quantity of data". The issue of data management is emerging. Information Lifecycle Management (ILM) means the whole procedure to align the business value of information with the most appropriate and cost-effective infrastructure from the time information is created through its final disposition. How to migrate data is one of the most important key points in ILM. Several independent files can migrates efficiently to the most appropriate storage medium in a hierarchical storage system under migration policy. But there are some problems if the migration targets are the data of multiple versions of the same source file Due to the native characteristic of snapshot, the snapshot data of the same source file is non-independent. In logic phase, the different versions data of the same source file can recognized as multiple files. But in the physical phase, the multiple files have shared partial data. Therefore, the improper migration problem will appears when the current migration policy applied to snapshot data. The improper migration will decrease the performance when read or write the newest version file. Duplicating the common part data can solve the improper migration problem but bring the new problem: redundancy data.
This paper sets forth the concept of Snapshot Lifecycle Management (SLM). SLM can manage the snapshot files in more appropriate and efficient way. In addition, this paper proposes the modified remote Copy-on-Write to improve performance under remote snapshot backup. Hence, the modified remote snapshot is more suitable for remote backup over WAN. SLM is not only fit for snapshot, it also can apply to all incremental backup mechanism and manage the mid-term backup storage system.

摘 要 i
Abstract ii
致 謝 iii
目 錄 iv
第一章 前 言 1
第二章 背景技術 8
2-1 ILM 8
2-2 Policy 14
2-3 Snapshot 17
第三章 Snapshot Lifecycle Management 25
3-1 SLM架構 25
3-2 快照管理 26
3-2-1 復原 26
3-2-2 快照資料管理 35
3-3 改良式遠端快照 38
第四章 實作與效能測試 40
4-1 實作分析 40
4-2 實驗環境 45
4-3 效能測試 46
第五章 結論與未來工作 48
參考文獻 50

[1] http://taiwan.cnet.com/enterprise/topic/0,2000062938,20105986,00.htm
[2]"Toward the Terabyte Era: A Look at New Developments in the Data-Storage World", Magnetic Media Information Services Commentator Volume 18: Issue 1 1st Quarter 2001
[3]Mandis Beigi, Murthy Devarakonda, Rohit Jain, Marc Kaplan, David Pease, Jim Rubas, Upendra Sharma, and Akshat Verma, “Policy-Based Information Management in a Large-scale File System”, Policies for Distributed Systems and Networks (POLICY’05)
[4]Murthy V. Devarakonda, “Method and Systems for Developing Data Life Cycle Policies”, U.S. Patent Application No. 20060059172
[5]Gauri Shah, Kaladhar Voruganti, Piyush Shivam, Maria del Mar Alvarez Rohena, “ACE: Classification for Information Lifecycle Management”, IBM Research Report (Feb. 22, 2006)
[6]Bruce Greenblatt, Claudia Chandra, Wen Cheng, “Techniques for Performing Policy Automated Operations”, U.S. Patent Application No. 20050033757
[7]SW Worth, Richard Saunders, "Backup & Recovery of Storage Networks Part II", Storage Networking World, 2003
[8]Application Note - Global Data and Storage Sharing, http://www.yottayotta.com/pdf/YottaYotta_Application_Note.pdf
[9]https://wealth.goldman.com/
[10]Wen-Shyang Hwang, Sheng-Hsiung Lin, Jui-Ting Chien, Jun-Yao Wang, and Shou-Jen Wey, "Design and Implementation of an iLVM for Remote Mirror", TANET 2005.
[11]International Technology Group, http://www.intertechnogroup.com
[12]Hewlett-Packard, http://www.hp.com.tw/
[13]StorageTek, http://www.storagetek.com.tw/
[14]Ching-Hung Lin, Wen-Shyang Hwang, Tsung-Lin Tsai, Chin-Pei Su, and Shou-Jen Wey, "A Novel Technique for Remote Snapshot Using Rollforward and Multiple-to-One Snapshot Management", TANET 2005.
[15]FalconStor, http://www.falconstor.com
[16]SNIA – Storage Networking Industry Association, http://www.snia.org
[17]Veritas, http://www.veritas.com/
[18]Snapshot, http://www.snia.org/dictionary
[19]Nik Simpson, "ILM: Tiered Storage & The Need for Classification", SNIA, http://www.snia.org/education/tutorials/2006/spring/data-management/ILM_Tiered_Storage_and_the_Need_for_Classification.pdf
[20]Michael Beck, Harald Bohme, Mirko Dziadzka, Ulrich Kunitz, Robert Magnus, Dirk Verworner, Claus Schroter, “Linuxe Kernel Programming”, Third Edition
[21]Alessandro Rubini, Jonathan Corbet, “Linux Device Drivers”, Second Edition
[22]Jonathan Corbet, Alessandro Rubini, Greg Kroah-Hartman, “Linux Device Drivers”, Third Edition
[23]David A Rusling, "The Linux Kernel", http://www.tldp.org/LDP/tlk/
[24]Remy Card, Theodore Ts'o, Stephen Tweedie, Design and Implementation of the Second Extended Filesystem, Proceedings of the First Dutch International Symposium on Linux