結合群聚技術之搜尋演算法於容錯系統之研究

The Study of Search Algorithm Combined Clustering Technique in Fault-Tolerant System

林鳳書(Feng-Shu Lin)

指導教授 ： 丁鏞

中原大學/工學院/機械工程研究所/碩士(2004年)

https://doi.org/10.6840/cycu200400016

摘要

本文主要是研究在容錯系統架構下，有關應用程式之資料備份技術。在資料備份中，最重要的就是即時監控技術，此技術會影響備份機制的資源使用情形。若以判斷式備份機制作為即時監控技術，發現它雖然可以避免不必要的備份工作，但卻十分佔用系統資源。本文提出利用Windows API的資料監控函式來取代判斷式備份機制，雖然監控函式確實能有效降低系統使用資源，不過當檔案異動頻繁時，有可能會造成檔案路徑的遺失。故本文另以多緒式平行搜尋演算法作檔案路徑之搜尋以解決路徑遺失問題。最後則根據檔案異動的歷史資料表以K-Means資料群聚技術將經常異動過的檔案路徑群聚出來，並結合多緒式平行搜尋演算法的方式去做資料搜尋，以達到更快速搜尋到檔案路徑之目的。

關鍵字

容錯系統 ； 搜尋演算法 ； 群聚技術

並列摘要

Data replication technique for fault tolerance system is developed in this article. Real-time monitor method is most important in data replication, which affects the usage of system resource. By means of the judgment method of the replication mechanism for real-time monitoring, it can avoid unnecessary task but occupy most system resource. Instead, in this study, the data monitor function of Windows API is used. Although the data monitor function is indeed able to reduce the use of system resource effectively, the file path may lose while files are changed frequently. Therefore, the multi-thread parallel search method is proposed to solve the problem of file path loss. The K-Means data-clustering method is used to find the changed files according to the file changed record table. Fast search of file path is achieved by the integration of the multi-thread method and the K-Means data clustering method.

並列關鍵字

Clustering Technique ； Fault-Tolerant System ； Search Algorithm

參考文獻

[14] 鄭武昇, “主從式容錯架構之檔案系統備份設計”, 中原大學機械工程研究所碩士論文, 2002年7月.

[2] Patterson, David A., Gibson, Garth, and Katz, Randy H., “A Case for Redundant Arrays of Inexpensive Disks” (RAID), In International Conference on Management of Data (SIGMOD), pages 109-116, June 1988.

[3] Hua, K.A., Khanh Vu, Ta-Hsiung Hu, “Improving RAID performance using a multibuffer technique”, Proc. 15th International Conference on Data Engineering, 23-26 Mar 1999, Pages 79 –86.

[5] K. Bell, A. Chien, M. Lauria, “A High-Performance Cluster Storage Server”, HPDC-11 2002. Proceedings. 11th IEEE International Symposium on, July 2002, pp. 311-320.

[6] Hu Jinfeng, Zhang Youhui, Zheng Weimin, “Orthogonal Replication Algorithm on Cluster Storage System”, Algorithms and Architectures for Parallel Processing, 2002. Proceedings. Fifth International Conference on, 23-25 Oct. 2002, pp. 422 –425.