雲端運算在許多科技企業的推動下發展相當迅速。雲端運算的能力通常是由分散式架構和虛擬化技術組成的，其中Kubernetes是屬於PaaS。Kubernetes的組態設定寫在YAML檔案中，所以需要學會看懂YAML檔案的內容，以及改寫程式碼，如果使用者有一個輔助生成YAML的軟體，對於Kubernetes的學習是有幫助。接著Kubernetes的預設調度會使用輪值查詢法，並假定每台節點主機性能相同，自動尋找資源足夠的主機，將工作服務佈署在可用的節點主機，但在某些情況下無法符合工作任務需要，或是主機因為處理太多繁雜的工作，導致資源分配太過分散。此外Kubernetes借由HPA(運算單位水平式自動擴張)功能，來處理類似網頁服務等一些高負載量工作，但是HPA啟動時會持續增加節點主機資源。本論文目的希望讓想學習Kubernetes的人，有一個YAML檔案設計軟體，透過中文解說的介面，方便使用者了解Kubernetes各項功能設定與參數調整。以及提出一個分類調度方法，結合權重值分配運算單位，可以將每個工作服務交付給適合的節點主機，另外利用Shell Scripts將HPA功能改進，重新設計演算方法，不只可以縮短擴張時間，更幫助減少運算資源的消耗。

Cloud computing has been developed rapidly by many technology companies. Cloud computing consists of decentralized architecture and virtualization technology. Kubernetes belongs to PaaS. Kubernetes’s configuration profiles are written in YAML files, so users need to study and program the contents of YAML files. Having an assistant software that automatically generates the YAML is helpful to learn Kubernetes. Kubernetes’s default scheduling is based on a round-robin method. It assumes that the performance of each host is the same, and automatically finds host resources, and deploys jobs to available hosts. Under certain situations, this scheduling approach does not really meet the needs of jobs. For example, some hosts handle too many complicated jobs and these make host resources scattered. In addition, Kubernetes uses HPA (Horizontal Pod Auto-scaler) to handle some high loading services such as web services, but HPA will continue to increase host resources when it starts.
The purpose of this thesis is to let users who want to learn Kubernetes have a YAML profile software using Chinese interface. It is convenient for users to understand various settings and adjustments of Kubernetes. We propose a classification scheduling that is combined with weighted allocation to make each job to be delivered to a suitable host. In addition, Shell Scripts are used to redesign the calculation method and improve HPA.

Chapter 1 前言 1
1.1 研究背景 1
1.2 研究動機與目的 3
1.3 論文架構 3
Chapter 2 文獻探討 5
2.1 Kubernetes發展現況 5
2.1.1 物聯網(Internet of Things) 5
2.1.2 人工智慧(Artificial Intelligence) 6
2.1.3 機械學習(Machine Learning) 6
2.1.4 邊緣計算(Edge Computing) 6
2.1.5 無伺服器計算(Serverless Computing) 7
2.1.6 雲端運算(Cloud computing) 7
2.2 應用技術介紹 8
2.2.1 虛擬化技術 8
2.2.2 Docker 8
2.2.3 Kubernetes 9
2.2.4 Prometheus Operator 11
2.2.5 Python 12
2.2.6 負載平衡排程法 13
2.2.7 決策樹分類法(監督式學習) 13
2.2.8 其他應用技術 14
Chapter 3 組態設定檔案生成設計 15
3.1 系統設計 15
3.2 Deployment 15
3.3 Service 18
3.4 HPA 20
3.5 Ingress 22
3.6 Volume 23
3.7 其他工具 26
Chapter 4 研究方法 27
4.1 Kubernetes部屬 27
4.2 測試系統 28
4.3 Kubernetes預設調度 29
4.4 分類調度法 30
Chapter 5 系統實驗與分析 37
5.1 實驗環境 37
5.2 實驗一：運算效能測試 38
5.2.1 Kubernetes預設調度 38
5.2.1 分類調度法 39
5.2.2 實驗結果 41
5.3 實驗二：網頁負載測試 41
5.3.1 POD數量與持續時間比較 42
5.3.2 CPU使用比較 42
5.3.3 記憶體使用比較 43
5.3.4 Network使用比較 44
5.3.5 實驗結果 45
5.4 實驗貢獻 46
Chapter 6 結論與未來工作 47
參考文獻 48

[1] 鄧鴻毅、林美佐、蔡文翊, 台灣雲端運算之現況與未來發展, 工研院資通所, 2017.
[2] 王宏仁, “IBM雲端平臺不惜砍掉重練，改用K8s打造現代化新架構,” 6 6 2019. [線上]. Available: https://www.ithome.com.tw/people/131015.
[3] 王宏仁, “VMware為何要用K8s重新打造vSphere？,” 10 9 2019. [線上]. Available: https://www.ithome.com.tw/news/132901.
[4] 王若樸, “K8s解決機器學習複雜的工作流程痛點，撐起電商AI服務,” 2 10 2019. [線上]. Available: https://www.ithome.com.tw/news/133300.
[5] Asad Javed, Keijo Heljanko, Andrea Buda and Kary Främling, “CEFIoT: A Fault-Tolerant IoT Architecture for Edge and Cloud,” IEEE, 2018.
[6] Corentin Dupont, Raffaele Giaffreda, Luca Capra, “Edge computing in IoT context: horizontal and vertical Linux container migration,” IEEE, 2017.
[7] Shapna Muralidharan, Gyuwon Song, Heedong Ko, “Monitoring and Managing IoT Applications in Smart Cities Using Kubernetes,” CLOUD COMPUTING 2019 : The Tenth International Conference on Cloud Computing, GRIDs, and Virtualization, 2019.
[8] Yuansong Qiao, Robert Nolan, Saul Gill, Guiming Fang, Brian Lee , “ThingNet: A Micro-Service based IoT Macro-Programming Platform over Edges and Cloud,” IEEE, 2018.
[9] 周冠言 姜勁帆 許景翔 陳政宇 朱金城, “容器化 TensorFlow 於異質 CPU/GPU 叢集之實作,” 台灣網際網路研討會, 2017.
[10] 江瑋哲 、劉明倫 、蔡順吉 、詹禮維 、郭千仞 、楊朝棟、賴傳霖 、郭嘉真 , “基於 Docker 與 Kubernetes 運算平台快速佈署,” 臺灣網際網路研討會 , 2018.
[11] 莊家雋 潘怡倫 陳惠珊 彭鈺翔 周家瑋 , “AITrain: 基於Kubernetes 之動態雲端資源分配教學平台,” 臺灣網際網路研討會 , 2018.
[12] Pei-Hsuan Tsai, Hua-Jun Hong, An-Chieh Cheng, and Cheng-Hsin Hsu, “Distributed Analytics in Fog Computing Platforms Using TensorFlow and Kubernetes,” IEEE, 2017.
[13] Swarnava Dey, Arijit Mukherjee, “Implementing Deep Learning and Inferencing on Fog and Edge Computing Systems,” Second International Workshop on Smart Edge Computing and Networking, 2018.
[14] Semir Sarajlic , Julien Chastang , Suresh Marru , Jeremy Fischer , Mike Lowe , “Scaling JupyterHub Using Kubernetes on Jetstream Cloud: Platform as a Service for Research and Educational Initiatives in the Atmospheric Sciences,” PEARC ‘18, 2018.
[15] W. Shi, “The Promise of Edge Computing,” IEEE, 2016.
[16] Yuzhou Huang, Kaiyu cai , Ran Zong , Yugang Mao , “Design and Implementation of an Edge Computing Platform Architecture Using Docker and Kubernetes for Machine Learning,” 2019 Association for Computing Machinery. , 2019.
[17] S. K. Mohanty, “Evaluation of Serverless Computing Frameworks Based on Kubernetes,” Aalto University School of Science Degree Programme of Computer Science and Engineering, 2018.
[18] Vladimir Podolskiy, Anshul Jindal and Michael Gerndt, “IaaS Reactive Autoscaling Performance Challenges,” 2018 IEEE 11th International Conference on Cloud Computing, 2018.
[19] B. Saha, “Green Computing,” International Journal of Computer Trends and Technology (IJCTT), 2014.
[20] 楊. 戴. 曹亞侖, Docker入門與實戰, 碁峰資訊.
[21] Xiao-Lan Xie,Peng Wang, Qi Wang, “The Performance Analysis of Docker and Rkt Based on Kubernetes,” 2017 13th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery, 2017.
[22] 龔正, Kubernetes 權威指南, 電子工業出版社.
[23] Chia-Chen Chang, Shun-Ren Yang, En-Hau Yeh, Phone Lin, Jeu-Yih Jeng, “A Kubernetes-Based Monitoring Platform for,” IEEE, 2017.
[24] Meina Song, Chengcheng Zhang and Haihong E, “An Auto Scaling System for API Gateway,” IEEE, 2018.
[25] M. Moilanen, “Deploying an application using Docker and Kubernetes,” Business Information Technology Oulu University of Applied Sciences, 2018.
[26] Leila Abdollahi Vayghan, Mohamed Aymen Saied, Maria Toeroe, Ferhat Khendek, “Deploying Microservice Based Applications with Kubernetes: Experiments and Lessons Learned,” 2018 IEEE 11th International Conference on Cloud Computing, 2018.
[27] Xin Guo, Jun Ni, “Improvement of Flexibility and Elasticity in Kubernetes-enhanced Docker Container Clustering,” The Ninth International Conference on Informatics and Computing for Science and Engineering, 2017.
[28] Scott Boag1, Parijat Dube1, Benjamin Herta1, Waldemar Hummer1, Vatche Ishakian1,2,, “Scalable Multi-Framework Multi-Tenant Lifecycle Management of Deep Learning Training Jobs,” 31st Conference on Neural Information Processing Systems, 2017.
[29] Towards a Full-Stack DevOps Environment (Platform-as-a-Service) for Cloud-Hosted Applications, TSINGHUA SCIENCE AND TECHNOLOGY, 2017.
[30] Amit Dua, Sahil Randive, Aditi Agarwal, Neeraj Kumar, “Efficient Load balancing to serve Heterogeneous”.Thapar University.
[31] 吳龍輝, Google御用Docker叢集管理Kubernetes, 佳魁資訊, 2018.