簡易檢索 / 詳目顯示

研究生: 林信夫
Hsin-Fu Lin
論文名稱: 應用於分散式系統的約定式檔案預取技術
Appointed Prefetching for Distributed File System of Thin-Client/Server Computing in WAN
指導教授: 黃冠寰
Hwang, Gwan-Hwan
學位類別: 碩士
Master
系所名稱: 資訊教育研究所
Graduate Institute of Information and Computer Education
論文出版年: 2004
畢業學年度: 92
語文別: 英文
論文頁數: 83
中文關鍵詞: thin-client/server分散式檔案系統約定式的檔案預取機制AFPL
英文關鍵詞: Thin-client/server computing model, distributed file systems, appointed file prefetching mechanism, AFPL
論文種類: 學術論文
相關次數: 點閱:160下載:1
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  • 隨著thin-client/server系統架構不斷改進、分散式檔案系統架構漸趨成熟及網路頻寬的增大,使用者可從任何一台thin-client存取遠端伺服主機所提供的服務,如同使用本機電腦一樣方便。黃冠寰博士提出的MAS TC/S系統是適用於廣域網路的thin-client/server架構,我們根據這個架構思考如何在廣域網路上解決因等候檔案在檔案伺服器與應用伺服器間傳輸所造成的延遲問題。此篇論文提出一種約定式的檔案預取機制來解決此問題。
    所謂約定式的檔案預取機制是由使用者運用Appointed File Prefetching Language (AFPL) 去設定預取指令,然後系統根據設定之指令將需要的檔案於適當的時間預取。預取指令包含兩個部份︰選取需要的檔案以及選定預取的方式。藉由從系統及past file access records所提供的檔案資訊,使用者可以選取適當的檔案作為預取的檔案集合,系統運用時間預取或事件預取來執行約定式預取指令︰時間預取是由使用者決定預取檔案集合該在何時傳送至特定應用伺服器的快取內,時間設定可為每日特定的時間、每週特定的時間、或是某個特定的時間。使用者並可設定一行事曆以配合時間預取運作;事件預取是由使用者決定在何種事件觸發後再執行預取指令將預取檔案集合傳送至目前使用者所使用的應用伺服器內,例如當使用者一登入,或是使用某個特定的程式之後。我們的實驗結果證明運用此一方法約可減少30%至90%的時間延遲,並可提高6%至18%的檔案命中率。

    With the continuous improvements of the technologies of the thin-client/server computing model, the maturity of distributed file system, and the rapid growth of the network bandwidth, a user with a thin-client device can connect to any application server via the network. Hwang et al. purposed the MAS TC/S architecture which targets on the application of thin-client/server computing model on wide area network (WAN). Based on the MAS TC/S architecture, this paper discusses how to reduce the file transformation time between the file server and the application server. In this paper, we propose an appointed file prefetching mechanism to solve this problem.
    With the appointed file prefetching mechanism, a user or system administrator can write programs which are based on the appointed file prefetching language (AFPL) to instruct the system to prefetch the required files on the specified times. The prefetching instruction on AFPL contains two parts. The first part is used to select the required files and the other is to specify the way for the system to perform prefetching. According to the information in the file system and obtained past file access records, a user can select the needed files as prefetching file set, and s/he can specify to perform time prefetching or event prefetching. The time prefetching is to do prefetching on specific times, like a daily work, a weekly work, or an occasional work. It also can cooperate with the user’s schedule. The event prefetching is to perform prefetching while specific events occur. The events can be the user logins or executing applications and so on. The experimental results show that the time of remote file fetch reduced by 30% to 90% and the hit ratio increased by 6% to 18% in most cases.

    LIST OF TABLES IV LIST OF FIGURES V 1. Introduction 6 1.1 Background and motive 6 1.2 Appointed prefetching mechanism 10 2. Related work 12 2.1 Distributed file system 12 2.2 Web prefetching 16 2.3 Automatic prefetching 18 3. Operating model of appointed file prefetching 23 3.1 Appointed file prefetching architecture 24 3.2 File related information 26 3.3 Appointed file prefetching program 27 3.4 Operating mechanism 27 3.5 Appointed file prefetching scenario 29 4. Appointed file prefetching language 32 4.1 Data structure used in AFPL 32 4.2 How to write an AFPL program 35 4.3 Implementation of AFPL 42 5. Simulation environment and experimental results 50 5.1 Simulation environment 50 5.2 Experiment 51 5.3 Experiment result analysis 60 6. Conclusion and future work 67 Reference 69 Appendix: .file_access_record example 72

    [1] Joel P. Kanter. Understanding Thin-Client/Server Computing. Microsoft Press, 1998.
    [2] Gwan-Hwan Hwang, San-Yih Hwang, Yun-Sheng Chen, Jia-Qing Li. MAS TC/S: Roaming Thin Clients in a Wide Area Network with Transparent Working Environments. Proceedings of Advanced Technologies and Applications for Next Generation Information Communication Networks, 2002.
    [3] Elizabeth Shriver, Christopher Small, Keith A. Smith. Why Does File System Prefetching Work? Proceeding of the USENIX Annual Technical Conference, 1999.
    [4] Yun-Sheng Chen, A Practical Approach for File Prefetching in Distributed File System, National Taiwan Normal University Graduate Institute of Computer Science & Information Engineering Master Thesis, 2003.
    [5] Hui Lei, Dan Duchamp. An Analytical Approach to File Prefetching. Proceeding of the USENIX Annual Technical Conference, 1997.
    [6] P. Cao, E. W. Felten, A. Karlin, K. Li. A Study of Integrated Prefetching and Caching Strategies. Proceeding of 1995 ACM SIGMETRICS.
    [7] R. H. Patterson, G. A. Gibson, E. Ginting, D. Stodolsky, J. Zelenka. Informed Prefetching and Caching. Proceeding of Fifteenth Symposium on Operating System Principles, ACM, 1995.
    [8] J. Griffioen and R.Appleton. Reducing File System Latency Using a Predictive Approach. In Proc. 1994 USENIX Summer Conf., pages 197-207, June 1994.
    [9] R. H. Patterson, G. A. Gibson, M. Satyanarayanan. A Status Report on Research in Transparent Informed Prefetching. Operating Systems Review, 1993.
    [10] T. M. Kroeger and D. D. E. Long. Predicting Future File-System Actions from Prior Events. In Proc. 1996 USENIX Annual Technical Conf., pages 319-328, January 1996.
    [11] F. Isaila. An Overview of File System Architectures. In “Algorithms for Memory Hierarchies”, Lecture Notes on Computer Science, Springer Verlag, 2003.
    [12] J. H. Howard. An Overview of the Andrew File System. In Proceedings of the USENIX Winter Technical Conference, page 23-26, 1988.
    [13] OpenAFS. http://www.openafs.org/.
    [14] NFS (Version 4). http://www.nfsv4.org
    [15] RFC 3010. http://www.faqs.org/rfcs/rfc3010.html.
    [16] RFC 1094. http://www.faqs.org/rfcs/rfc1094.html.
    [17] RFC 1813. http://www.faqs.org/rfcs/rfc1813.html.
    [18] Coda File System. http://www.coda.cs.cmu.edu/.
    [19] Peter J. Braam. The Coda Distributed File System. Linux Journal, #50, June 1998, p 46-51.
    [20] Global File System. http://www.sistina.com/products_gfs.htm.
    [21] InterMezzo. http://www.inter-mezzo.org/.
    [22] The Sprite Opening System. http://www.cs.berkeley.edu/projects/sprite/sprite.html.
    [23] M. Nelson, B. Welch, and J. Ousterhout. Caching in the Sprite network file system. ACM Transactions on Computer Systems, 6(1):134-154, February 1988.
    [24] Serverless File System called xFS. http://now.cs.berkeley.edu/Xfs/xfs.html.
    [25] A. Nanopoulos, D. Katsaros, Y. Manolopoulos. Effective Prediction of Web-user Accesses: A Data Mining. Proceeding of the WEBKDD 2001.
    [26] A. Nanopoulos, D. Katsaros, Y. Manolopoulos. Exploiting Web Log Mining for Web Cache Enhancement. Lecture Notes on Artificial Intelligence (LNAI), Springer-Verlag, vol. 2356, pp. 68-87, 2002.
    [27] Cheng-Zhong Xu and Tamer I. Ibrahim. Towards Semantics-Based Prefetching to Reduce Web Access Latency. Proc. of the 2003 Int’l Symp. On Applications and the Internet (SAINT’2003) pages 318-325, January 2003.
    [28] Kroeger KM, Long DDE. Predicting File System Actions from Prior Events. Proceedings of the USENIX 1996.
    [29] Kroeger KM, Long DDE. The case for efficient file access pattern modeling. In Proceedings of the Seventh Workshop on Hot Topics in Operating Systems (HotOS-VII), IEEE, March 1999.
    [30] Kroeger KM, Long DDE. Design and Implementation of a Predictive File Prefetching Algorithm. In Proceedings of the 2001 USENIX Annual Technical Conference.
    [31] G. Coulouris, J. Dollimore, T. Kindberg. Distributed systems concepts and design, 3rd ed. (Addison-Wesley, 2001), 67 page.

    QR CODE