隨著雲端運算的興起，分散式系統的重要性逐漸受到重視，大量複雜的系統與服務逐漸成為企業的常態。而容器技術的普及，又為新一波軟體部署與管理方式帶來了前所未有的革命，其中以Docker和Kubernetes 等平台的出現，讓微服務的設計概念得以落實。微服務的設計思維不僅能夠降低系統中服務與服務之間的耦合性，也能夠達成跨程式語言的開發模式。傳統資料存取模式(增刪修改)，在面對大量使用者同時寫入的情況下，經常造成單一失效及效能問題。使用命令與查詢責任分離模式 (CQRS, Command Query Responsibility Segregation) 與事件溯源 (ES, Event Sourcing)架構，由軟體層面增加了系統的可擴張性，然而這些機制在基礎設施層面上目前仍欠缺完整的錯誤偵測與回復功能，且面對複數使用者同時對資料庫進行改寫時，亦會遇到一致性的問題。本論文採用CQRS/Event Sourcing架構的系統，設計一個完整且自主的錯誤偵測與回復機制，亦可處理並行讀寫的問題，並透過實作與數據分析驗證此設計機制的可行性。

As the Cloud computing raise rapidly, distributed system gradually attach great importance to a lot of software engineers around the world. Having a great range of complex business’s logic system and services become enterprise’s new normal condition. With the universal of container technology, the new wave of unprecedented revolution has spread into all kinds of software deployments and software managements. The appearance of Kubernetes and Docker platforms makes Microservice pattern have the chance to implement in enterprise’s business system. The design concept of Microservice pattern not only can reduce the coupling between service and service but also can allow user to develop system in multi-platform environment. In the traditional method for data access(CRUD), when a lot of users’ update data simultaneously, this will often cause single failure and low performance. Using Command Query Responsibility Segregation(CQRS) and Event Sourcing can enhance the scalability of system in software level. However, these mechanisms in infrastructure aspect still lack of complete failure detection and failure recovery. Additionally, when facing a huge amount of users update the database at the same time, will also encounter consistency problem. This essay will adopt the pattern of CQRS/Event Sourcing Architecture to design a complete failure detection and automatic failure recovery, this also can solve the problem of data access in parallel. We verify the availability of this system through implementation and data analysis.

摘要 I
第一章 緒論 - 1 -
1.1研究背景 - 1 -
1.2研究動機 - 3 -
1.3研究貢獻 - 4 -
1.4 論文架構 - 4 -
第二章 技術背景與相關研究 - 5 -
2.1 命令與查詢責任隔離(COMMAND QUERY RESPONSIBILITY SEGREGATION) - 5 -
2.2 EVENT SOURCING - 7 -
2.3 CIRCUIT BREAKER - 9 -
2.4 MESSAGE-ORIENTED MIDDLEWARE(MOM) - 12 -
2.5 RAFT演算法 - 14 -
2.6相關研究 - 19 -
第三章 系統設計 - 21 -
3.1系統架構說明 - 21 -
3.1.1 系統架構 - 22 -
3.1.2 Service端 & Watch Dog - 24 -
3.1.3 Event Listener, Event Store, 和Recovery Service機制 - 25 -
3.1.4 CQRS運作流程解說 - 26 -
3.2 強化集中式監控系統(WATCHDOG) - 28 -
3.2.1 WatchDog 機制解說 - 28 -
3.3 增強因單一失效而造成崩潰的分散式系統可用性(CIRCUIT BREAKER) - 31 -
3.3.1 Circuit Breaker - 32 -
3.4 排除EVENT SOURCING系統交易中可能出現的競爭危害 - 35 -
3.4.1 交易未寫入情況(Race condition problem) - 35 -
3.4.2 交易retry機制 - 36 -
3.5 錯誤偵測及錯誤回復機制 - 37 -
3.5.1 Service端未回應(Service A 失效) - 37 -
3.5.2 Service 重啟機制 (Failure recovery) - 38 -
3.6 使用場景 - 40 -
第四章 實驗測試與討論 - 42 -
4.1 實驗總覽 - 42 -
4.2 系統錯誤回復效能 - 42 -
4.3 WATCHDOG 共識演算法恢復效能 - 44 -
4.4 RACE CONDITION PROBLEM 解決效能 - 46 -
4.5 討論 - 48 -
第五章 結論 - 50 -
參考文獻 - 52 -

[1] C. Richardson, Microservices Patterns
With examples in Java. Published 2017 by Manning Publications, October 2018.
[2] M. T. Nygard, Release It!: Design and Deploy Production-Ready Software. Pragmatic Bookshelf, 2018.
[3] Z. Long, "Improvement and implementation of a high performance CQRS architecture," in 2017 International Conference on Robots & Intelligent System (ICRIS), 2017: IEEE, pp. 170-173.
[4] B. Meyer, Object-oriented software construction. 1997.
[5] R. Martin, Agile Software Development, Principles, Patterns, and Practices. Prentice Hall, 2002.
[6] G. Young, CQRS Documents. 2010.
[7] M. Fowler. "Event Sourcing." https://martinfowler.com/eaaDev/EventSourcing.html (accessed.
[8] M. T. Nygard, Release It!: Design and Deploy Production-Ready Software. 2007.
[9] M. J. Fischer, N. A. Lynch, and M. S. Paterson, "Impossibility of distributed consensus with one faulty process," Journal of the ACM (JACM), vol. 32, no. 2, pp. 374-382, 1985.
[10] S. Gilbert and N. Lynch, "Perspectives on the CAP Theorem," Computer, vol. 45, no. 2, pp. 30-36, 2012.
[11] D. Ongaro and J. Ousterhout, "In search of an understandable consensus algorithm (extended version)," ed: Tech Report. May, 2014. http://ramcloud. stanford. edu/Raft. pdf, 2013.
[12] M. Overeem, M. Spoor, and S. Jansen, "The dark side of event sourcing: Managing data conversion," in 2017 IEEE 24th International Conference on Software Analysis, Evolution and Reengineering (SANER), 2017: IEEE, pp. 193-204.
[13] D. Meißner, B. Erb, and F. Kargl, "Performance Engineering in Distributed Event-sourced Systems," presented at the Proceedings of the 12th ACM International Conference on Distributed and Event-based Systems, Hamilton, New Zealand, 2018. [Online]. Available: https://doi.org/10.1145/3210284.3219770.
[14] J. Rybicki, "Application of Event Sourcing in Research Data Management," 2018 : The Fourth International Conference on Big Data, Small Data, Linked Data and Open Data, pp. 22-26, 2018.
[15] S. Han and J.-i. Choi, "V2X-Based Event Acquisition and Reproduction Architecture with Event-Sourcing," in Proceedings of 2020 the 6th International Conference on Computing and Data Engineering, 2020, pp. 164-167.
[16] Y. Zhong, W. Li, and J. Wang, "Using Event Sourcing and CQRS to Build a High Performance Point Trading System," in Proceedings of the 2019 5th International Conference on E-Business and Applications, 2019, pp. 16-19.
[17] A. Bellemare, Building Event-Driven Microservices: Leveraging Organizational Data at Scale. O'Reilly, 2020-07.
[18] M. Barnkob and J. Krukow, "Event Sourcing and Command Query Responsibility Segregation Reliability Properties," Computer Science University of Aarhus.—2018.—C, pp. 21-39, 2018.