臺灣博碩士論文加值系統

近年來，筆記型電腦、手機、各種 IoT等聯網設備的數量和多樣性日益增加。這些設備引入了大量的網路流量，使得網絡管理比以往更加的複雜。因此 AT＆T和ON.LAB (ONF) 及其他幾家主要的電信商、研究機構，一起推出了一個新的專案，叫做Central Office Re-architected as a Data Center ，簡稱為 CORD。
CORD結合了軟體定義網絡 (SDN)、網絡功能虛擬化 (NFV) 和雲端技術的高度彈性的特性，將資料中心的經濟效益和敏捷性帶進電信機房。而OpenCORD 是一套 CORD的開放實做參考，由商業化的伺服器、白牌網路交換器和開放原始碼軟體 (例如OpenStack、ONOS、XOS) 建置而成。然而，OpenCORD存在三個主要的問題：硬體需求昂貴、自動化工作不可靠，工具鏈複雜、功能重疊。這些問題使得新手難以參與 OpenCORD 這個專案。
在本文中，我們首先提出一種新的 CORD基礎架構，使用容器 (Container) 來取代虛擬機 (Virtual Machine) 提供服務，也因此我們不再需要使用 OpenStack，進而減少了對硬體資源的需求。並且我們引進了軟體測試、持續性整合以及持續交付方法，以強化交付及部署CORD基礎架構的可靠性。此外，我們所選擇的自動測試、持續整合、持續交付等工具，也都比 OpenCORD 中所使用到的工具更佳輕量化、對使用者友善。
實驗結果顯示，我們的CORD基礎架構需要更少的硬體資源，而且比 OpenCORD更加的可靠、建置所花費的時間更短。事實上，我們的 CORD 基礎架構自動部署過程每一次都是成功的，這是因為我們基礎架構每一次的變更都會自動的觸發整合測試，以確保被提交的版本的正確性。

In recent years, the number and the diversity of Internet devices, such as laptops, mobile phones, and various IoT devices, have been growing tremendously. These devices introduce a huge amount and variety of traffics that make network management more complicate than ever before. Therefore, AT&T and ON.LAB (ONF), together with several other major operators or research institutes, have launched a new project, Central Office Re-architected as a Data Center (CORD).
CORD combines Software Defined Networking (SDN), Network Function Virtualization (NFV) and the elasticity of commodity clouds to bring datacenter economics and cloud agility to the Telco Central Office. OpenCORD is an open reference implementation of CORD that can be built from commodity servers, white-box switches, and open source software (e.g., OpenStack, ONOS, XOS). However, OpenCORD possesses three major problems: expensive hardware requirement, unreliable automation, and complex and redundant toolchain. These problems make it very difficult for newbies to participate in the OpenCORD community.
In this thesis, we first propose a new CORD infrastructure that uses containers, instead of VMs, so that OpenStack is no longer required, and thus decrease the demand of hardware resources. We then introduce Software Testing, Continuous Integration, and Continuous Delivery, to enforce the reliability of the delivery and deployment of our CORD infrastructure. Furthermore, the tools we choose for the automatic testing, integration, and delivery are much lighter and more friendly than the ones used in OpenCORD.
The experimental result show that our CORD infrastructure requires fewer hardware resources and is more reliable with a shorter elapsed time than OpenCORD. Indeed, the automatic deployment process of our CORD infrastructure always succeeds because every change in our CORD infrastructure will automatically trigger an integration test to ensure the correctness of the committed version.

摘要 i
Abstract iii
誌 謝 v
目錄 vi
圖目錄 viii
表目錄 ix
第1章 Introduction 1
1.1 Motivation 1
1.2 Objective 2
1.3 Synopsis 3
第2章 Background 3
2.1 CORD - Central Office Re-architected as a Datacenter 3
2.2 OpenCORD - an Open Reference Implementation of CORD 5
2.3 CORD in a Box 6
2.4 Version Control System: Git & GitLab 7
2.5 Continuous Integration (CI) & GitLab CI 9
2.6 Configuration management: Ansible 10
2.7 Shell Script 10
第3章 Related Work 12
3.1 Continous Integration in OpenCORD 12
3.2 Problems in OpenCORD 12
3.2.1 Low quality in automation reliability 12
3.2.2 High-end hardware resource requirement 13
3.2.3 The complexity and difficulty 13
第4章 A Reliable CORD infrastructure using CI/CD technology 15
4.1 Design Issues 15
4.2 Architecture 16
4.3 Build Process 18
4.4 More robust and reliable software testing & Continuous Integration 20
4.5 Reduce the resource consumption 24
4.6 Simplify the toolchains and dependencies 25
第5章 Experimental Evaluation 27
5.1 Experieiment Environment 27
5.2 Installation time spent and success rate 28
第6章 Conclusion and Future Work 30
6.1 Conclusion 30
6.2 Future Work 30

[1] Ansible Documentation, https://docs.ansible.com/ansible/latest/index.html
[2] GitLab Documentation, https://docs.gitlab.com/
[3] Advanced Bash-Scripting Guide - The Linux Documentation Project, http://www.tldp.org/LDP/abs/abs-guide.pdf
[4] Pro Git, https://git-scm.com/book/en/v2
[5] OpenCORD Wiki, https://wiki.opencord.org/
[6] Libvirt, https://libvirt.org/
[7] ShellCheck wiki, https://github.com/koalaman/shellcheck/wiki/
[8] Docker User Guide, https://docs.docker.com/engine/userguide/
[9] Larry Peterson, Ali Al-Shabibi, Tom Anshutz, Scott Baker, Andy Bavier, Saurav Das, Jonathan Hart, Guru Palukar, William Snow, "Central office re-architected as a data center", IEEE Communications Magazine, October 2016
[10] Christof Ebert, Gorka Gallardo, Josune Hernantes, Nicolas Serrano, “DevOps”, IEEE Journals & Magazine Volume: 33 Issue: 3, April 2016
[11] Nishant Kumar Singh, Sanjeev Thakurt, Himanshu Chaurasiyal, Himanshu Nagdev, “Automated provisioning of application in IAAS cloud using Ansible configuration management”, Next Generation Computing Technologies (NGCT), Septemper 2015
[12] Celal Ziftci, Jim Reardon, “Who broke the build? Automatically identifying changes that induce test failures in continuous integration at Google Scale”, 2017 IEEE/ACM 39th International Conference on Software Engineering: Software Engineering in Practice Track (ICSE-SEIP), August 2017
[13] Mohammad Masudur Rahman, Chanchal K. Roy, “Impact of Continuous Integration on Code Reviews”, 2017 IEEE/ACM 14th International Conference on Mining Software Repositories (MSR), May 2017
[14] Fiorella Zampetti, Simone Scalabrino, Rocco Oliveto, Gerardo Canfora, Massimiliano Di Penta, “How Open Source Projects Use Static Code Analysis Tools in Continuous Integration Pipelines”, 2017 IEEE/ACM 14th International Conference on Mining Software Repositories (MSR), May 2017
[15] Md Rakibul Islam, Minhaz F. Zibran, “Insights into Continuous Integration Build Failures”, 2017 IEEE/ACM 14th International Conference on Mining Software Repositories (MSR), May 2017
[16] Mojtaba Shahin, Muhammad Ali Babar, Liming Zhu, “Continuous Integration, Delivery and Deployment: A Systematic Review on Approaches, Tools, Challenges and Practices”, IEEE Access Volume: 5, March 2017
[17] Torvald Mårtensson, Daniel Ståhl, Jan Bosch, “Continuous Integration Impediments in Large-Scale Industry Projects”, 2017 IEEE International Conference on Software Architecture (ICSA), May 2017