臺灣博碩士論文加值系統

摘要
於格網計算網路或分散式系統中，可以使用MPI訊息傳遞，來執行分散式異質平台中的平行應用程式。然而，藉由容錯機制，確保系統的可靠度，是非常重要的一環。因此，在這篇論文中，我們將提出一個在格網計算環境中，以MPI為基礎的容錯中介層─FATOM─「Fault Tolerance Middleware」。在這個中介層中，我們提供了一個非常簡單的方法，經由加上編譯器的註釋，來達成容錯的效果。如此一來，即使不重新撰寫整個應用程式，仍能達到容錯的效果。此外，透過FATOM這個容錯機制，應用程式將更有效率、更可靠得執行。在FATOM的測試方面，我們在使用FATOM與不使用FATOM的情況下，分別以最大-最小搜尋樹演算法，測試黑白棋的遊戲程式。實驗結果顯示，FATOM不會花費太多額外的資源，並且在有些時候，比沒有容錯機制的MPI程式有更佳的執行效率。

Abstract
In grid computing networks and distributed systems, parallel applications can run in distributed heterogeneous sites using MPI message-passing. It is important to enhance the reliability of the system with fault tolerance. In this paper, we discuss an MPI based middleware, FATOM (FAult TOlerance Middleware), for grid computing applications. The middleware provides an easy way for achieving fault tolerance by adding compiler directives, but not rewriting application programs. Through the fault tolerance capability, applications may be executed more efficiently and more reliably. We test programs of the minimum-maximum search tree of Othello game with and without FATOM. The experimental results indicate that FATOMdoes not cost much overhead and, sometimes, runs faster then the MPI program without fault tolerance.

Contents
誌謝 ..........................i
摘要 ..........................ii
Abstract .....................iii
1 Introduction ...........................................................1
2 Related Work .........................................................3
2.1 Overview of the MPI ...........................................3
2.2 Overview of the Grid Computing ........................4
2.3 Overview of the Globus Toolkit ..........................5
3 The Fault Tolerance Middleware: FATOM .............8
4 Program Implementation with FATOM ..................13
5 Experiments and Results ........................................17
6 Conclusions ............................................................21
Bibliography ..............................................................22

Bibliography
[1] F. Berman, G. Fox, and A. J. G. Hey, editors. Grid Computing: Making the Global Infrastructure a Reality. John Wiley, 2003.
[2] A. Bouteiller, P. Lemarinier, G. Krawezik, and F. Cappello. Coordinated checkpoint versus message log for fault tolerant MPI. In Proceedings of IEEE International Conference on Cluster Computing, pages 242–250, 2003.
[3] M. Buro. How machines have learned to play Othello. IEEE Intelligent Systems, 14(6):12–14, 1999.
[4] M. S. Campbell and T. A. Marsland. A comparison of minimax tree search algorithms. Artificial Intelligence, 20:347–367, 1983.
[5] E. N. Elnozahy and W. Zwaenepoel. On the use and implementation of message logging. In Proceedings of the 24th International Symposium on Fault-Tolerant Computing, pages 298–317, 1994.
[6] I. Foster. Designing and Building Parallel Programs: Concepts and Tools for Parallel Software Engineering. Addison Wesley, 1995.
[7] I. Foster. What is the Grid? A three point checklist. Daily News and Information for the Global Grid Community, 1(6), 2002. http://www.gridtoday.com/02/0722/100136.html.
[8] I. Foster and C. Kesselman. Globus: A metacomputing infrastructure toolkit. Inernational Journal of Supercomputer Applications, 11(2):115–128, 1997.
http://www.globus.org/.
[9] I. Foster and C. Kesselman, editors. The Grid 2: Blueprint for a New Computing Infrastructure. Morgan Kaufmann, 2003.
[10] W. Gropp, E. Lusk, N. Doss, and A. Skjellum. A high-performance, portable implementation of the MPI message passing interface standard. Parallel Computing, 22(6):789–828, 1996.
[11] N. Karonis, B. Toonen, and I. Foster. MPICH-G2: A grid-enabled implementation of the message passing interface. Journal of Parallel and Distributed Computing, 63(5):551–563, 2003.
[12] S. S. Vadhiyar and J. J. Dongarra. SRS - A framework for developing malleable and migratable parallel applications for distributed systems. Parallel Processing Letters, 13(2):291–312, 2003.