摘要
本论文研究的主要内容是运用数学建模的方法,对高可用性系统的设计进行量化分析。主要研究了高可用性系统设计中的量化分析方法。针对高可用性系统的设计提出了一个量化分析的框架:以UML建模为基础,从UML转换到随机Petri网(SPN)的建模方案。将提出一个UML转换SPN的算法,并基于SPN,运用与SPN同构的马尔可夫(Markov)随机过程来进行量化数值分析,得出系统设计的性能参数,并且利用这些性能参数来优化系统设计。
随着计算机应用的不断深入,企业对计算机系统的依赖程度也日渐增加。尤其在一些关键行业的关键应用上,应用的后台核心系统是否具有保护业务关键数据的完整和维持应用程序的高可用性的能力,已经成为影响一个公司业务成败的关键因素。高可用性是目前对于计算机系统的一个重要的非功能性需求。
高可用性系统的设计比一般的软件设计要复杂得多。如何在设计阶段来预测目标系统的量化参数是一个问题。这就是本论文的出发点。
在高可用性系统的设计阶段引入量化分析的方法,有着三方面的优点:一是对于系统的投资者,可以在快速知道目标系统的性能投资,可以在项目的初期阶段对投资的回报作出接价;二是对于高可用性系统的设计者来说,可以根据量化分析的结果来合理选择合适的技术方案和改进系统的设计方案;三是对于软件的质量保证(Quality Assurance)来说,可以把量化分析的结果和实际的测试结果进行分析,进一步验证系统的实现。
The primary research scope of this thesis is quantitative analysis to high availability system design using modeling method. A quantitative analysis framework for estimating the high available system design has been presented in this thesis. The steps for quantitative analysis are like these: using Unified Modeling Language (UML) to create a model of system design; then based on a conversion algorithm, translate the UML statechart graph into Stochastic Petri Net (SPN); after that, using a Markov Chain that is isomorphic to SPN to do the performance evaluation and get the performance benchmarks. Such data can be used for refining the system design.With the enormous computational power, enterprises become more dependent on the computer systems, especially for some critical industry application. It is quite important to protect the consistence of the data and maintain the high availability of application. How to provide a High Available Service becomes a popular non-functional requirement for a computer system.However, the design to a High Available system is more complicate than a common computer system. How to predict the quantitative parameters of target system is the main problem that will be tried to solve.There are three benefits by bringing quantitative analysis method to high available system design. For the system investor, this method can provide the interesting number which can be used for return assessment at the beginning of system design. For the system architect, this method can assist the decision to technical choice and refine the system design. For the software quality assurance (SQA), this method can help them to verify the system implemtation.
引文
[01] Ahmad, □luster Computing: A Glance at Recent Events? IEEE Concurrency, January-March 2000, vol. 8, no. 1, pp. 67-69.
[02] J. Muppala, G. Ciardo, and K.S. Trivedi, □tochastic Reward Nets for Reliability Prediction? Communications in Reliability, Maintainability and Serviceability: An International Journal published by SAE International, July 1994, Vol. 1, No. 2, pp. 9-20.
[03] M. Ajmone Marsan, G. Conte, and G. Balbo. □A class of generalized stochastic Petri nets for the performance evaluation of multiprocessor systems□. ACM Trans. Comput. Syst., 2(2): 93 {122, May 1984.
[04] C.M. Woodside, J.E. Neilson, D.C. Petriu, and S. Majumdar. □The stochastic rendezvous network model for performance of synchronous client-server-like distributed software? IEEE Transactions on Computers, 1995, 44(1): 20-34
[05] K. Trivedi □robability and Statistics with Reliability, Queuing and Computer Science Applications? Prentice-Hall
[06] Z. Cheng, and X. Yang, □High Available Software Architecture based on Cluster Technology? Proceedings of IEEE TENCON 2002, 327-330
[07] Kolesnikov and B. Hatch, □Building Linux Virtual Private Networks? New Riders Publishing, February 2002.
[08] D.I. Heiman, N. Mittal, and K.S. Trivedi, □vailability and Reliability Modeling for Computing Systems? Advances in Computers, 1990, Vol 31, pp. 175-233.
[09] M. Malhotra and K.S. Trivedi, □Dependability Modeling Using Petri-Nets? IEEE Transaction on Reliability, Sept. 1995, Vol. 44, No. 3, pp.428-440.
[10] X. Yang and W. Chen, □Quantitative Availability Prediction for a High Availability Messaging System? Proceedings of IEEE TENCON 2005.
[11] J. Rurnbaugh, I. Jacobson and G. Booch, □The Unified Modeling Language Reference Manual□, Addison-Wesley, ISBN: 020130998X
[12] R.A. Maxion and R.T. Olszewski. □etec tion and Discrimination of Injected Network Faults? Proc. 23 rd Int□I Symp. Fault-Tolerant Computing, 1993.
[13] M. Chen, A. Zheng, J. Lloyd, M. Jordan, and E. Brewer. Failure diagnosis using decision trees. In Proc. Intl. Conference on Autonomic Computing, New York, NY, 2004
[14] Storage Networking Solutions, □Achieving High Availability Objectives, □ www.cnt.com, 2003
[15] E. Yourdon and L. Constantine, □Structured Design: Fundamentals of a Discipline of Computer Program and Systems Design□, Yourdon Press, Englewood Cliffs, N.J., 1979
[16] P. Ward, and S. Mellor, □Structured Development for Real-Time Systems: Introduction and Tools□, Yourdon Press, Englewood Cliffs, N.J., 1985
[17] S. Shlaer and S. Mellor, □bject-Oriented Systems Analysis: Modeling the World in Data□, Yourdon Press, Englewood Cliffs, N.J., 1988
[18] P. Coad and E. Yourdon, □Object-Oriented Analysis, 2nd ed. □, Yourdon Press, Englewood Cliffs, N.J., 1991
[19] G. Booch, □Object-Oriented Analysis and Design with Applications, 1st ed.? Benjamin/Cummings, Redwood city, Calif., 1991
[20] J. Rumbaugh, M. Blaha, W. Premerlani, F. Eddy and W. Lorensen, □bject-Oriented Modeling and Design? Prentice Hall, 1991
[21] R. Wirfs-Brock, B. Wilkerson and L. Wiener, □esigning Object-Oriented Softer□, Prentice Hall, 1990
[22] A. Goldberg and D. Robson, □Smalltalk-80: The Language and Its Implementation? Addison-Wesley, 1983
[23] B. Cox, □bject-Oriented Programming: An Evolutionary Approach? Addison-Wesley, 1986
[24] B. Meyer, □Object-Oriented Software Construction□, Prentice Hall, 1988
[25] I. Jacobson, M. Christerson, P. Jonsson and G. Overgaard, □Object Oriented Software Engineering: A Use Case Driven Approach? Addison-Wesley, 1992
[26] D. Coleman, P. Arnold, S. Bodoff, C. Dollin, H. Gilchrist, F. Hayes and P. Jeremaes, □bject-Oriented Development: The Fusion Method? Prentice Hall, 1994
[27] G. Booch, I. Jacobson, and J. Rumbaugh, □The Unified Modeling Language User Guide? Addison Wesley 1999.
[28] V. Cortellessa and R. Mirandola, □Deriving a Queueing Network Based Performance Model from UML Diagrams? Proc. 2nd Int. Workshop on Software and Performance (WOSP), Ottawa, Canada, 58-69, 2000.
[29] V. Cortellessa and R. Mirandola, □UML Based Performance Modeling of Distributed Systems□, in: A. Evans, S. Kent, B. Selic, (Eds.), 3rd Int. Conf. on the Unified Modeling Language, York, UK, LNCS 1939, 178-193, Springer, 2000.
[30] C. Lindemann, □erformance Modeling with Deterministic and Stochastic Petri Nets? John Wiley & Sons, 1998.
[31] D. Gross and C. Harris. □Fundamentals of Queueing Theory□. Wiley, 3rd edition, 1998.
[32] P. King and R. Pooley. □Using UML to derive stochastic Petri net models□. In Proceedings of the 15th UK Performance Engineering Workshop, pages 45□56, Bristol, UK, July 1999.
[33] Object Management Group, □OMG Unified Modeling Language Specification□, http://www.omg.org.
[34] J. Muppala, M. Malhotra, K. S. Trivedi, □arkov Dependability Models of Complex Systems: Analysis Techniques, Reliability and Maintenance of Complex Systems□, S.Ozekici (ed.), pp. 442-486, Springer-Verlg, Berlin, 1996.
[35] Choi, H., Kulkami, V. G. and Trivedi, K. S. (1994) Markov regenerative stochastic Petri nets. Perform. Eval., 20, 337?57
[36] Hoon Choi, V.G. Kulkami, and K. Trivedi. Markov regenerative stochastic Petri nets. Performance Evaluation, 20: 337-357, 1994.
[37] T. Murata, □Petri Nets: Properties, Analysis and Applications? Proceedings of the IEEE, Vol. 77, No 4, April, 1989, pp. 541-580.
[38] Petri C A. Kommunkation mit automation. Bonn: Institut fur Lnstrum Entelle Mathematik, Schriften des ⅡM Nr. 3, 1962. Aslo, English translation, Communication with Automata. New York: Griffiss Air Force Base, Tech. Rep. RADC-TR-65-377, 1966, 1(1)
[39] Peterson J L. □etri net theory and the modeling of systems? Englewood Cliffs, NJ: Prentice-Hall
[40] Molly M K. □Discrete time Stochastic Petri Nets? IEEE Trans. Software Eng. 1985, SE-11(4): 417～423
