English  |  正體中文  |  简体中文  |  全文笔数/总笔数 : 80990/80990 (100%)
造访人次 : 40889750      在线人数 : 1380
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜寻范围 查询小技巧:
  • 您可在西文检索词汇前后加上"双引号",以获取较精准的检索结果
  • 若欲以作者姓名搜寻,建议至进阶搜寻限定作者字段,可获得较完整数据
  • 进阶搜寻


    jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/89933


    题名: 設計與實作結合Kubernetes應用之多執行緒連線負載平衡器;Design and Development of a Multi-threaded Connection Load Balancer for Kubernetes Application
    作者: 魏宏宇;Yu, Wei Hong
    贡献者: 資訊管理學系
    关键词: 負載平衡器;入口;多執行緒;Kubernetes;Ingress;multi-thread
    日期: 2022-09-22
    上传时间: 2022-10-04 12:05:09 (UTC+8)
    出版者: 國立中央大學
    摘要: 雲端叢集管理平台Kubernetes近年大受歡迎,逐漸為許多管理者所採用,各種應用隨之興起,乘著Kubernetes的風潮無伺服器 (Serverless) 的相關應用也逐漸出現,開源專案百花齊放,有各式各樣的無伺服器專案開始受到矚目。因此開始有學者針對無伺服器的效能進行研究,從過往相關文獻發現在特定的無伺服器專案中出現了效能低落的問題,這是由於其負載平衡器造成了系統的瓶頸。為了使負載平衡器能夠同時負擔巨大的流量,因此多執行緒的負載平衡器有望成為其解決方案的關鍵,而本論文中將針對其多執行緒對負載平衡器效能的影響來進行評估。
    為進行研究,本論文針對Kubernetes設計並實作了一個多執行緒的負載平衡器,並對於不同執行緒數量的效能進行觀察,實驗結果發現無限制地允許執行緒開啟無法帶來最好的效能表現,由於本論文所建置之負載平衡器連線之建立必須基於Kubernetes的API來開啟直通至 Pod 之專用Socket的緣故,負載平衡器一次性接受了大量來自測試客戶端的流量之後,同時向Kubernetes API要求建立Socket,此舉將造成了系統的壅塞並導致整體效能低下,進而使得整體連線回應時間不理想。因此本論文另外針對有限制數量的執行緒進行實驗,希望能夠從不同數量的執行緒了解其最佳化之執行緒數量避免連線壅塞的問題。從實驗結果中可以找出測試環境中Pod數量、連線要求、執行緒達到最佳執行效能的配置組合,在該配置之下可以得到最大的改善,並且比無限制執行緒數量的表現還好。
    ;The Kubernetes platform which aims to manage the cloud cluster platform is emergence with a lot of adoptions. Various kinds of application came out followed with this trend. There are o countless open-source projects published, many Serverless projects become the spotlight. There also being targeted as research topic. A Serverless platform are found with low performance issue in the research. The Load balancer are accused as the bottleneck of the system. To make a Load balancer able to load multiple flows from different sources. Multithreads are essential of Load balancer. In this thesis we are evaluating the performance of the Multithreaded Load balancer.
    In this thesis we design and implemented a multithreaded Load balancer and done an observation on its performance with different thread numbers. We found that non-limited thread expansion will not bring the best performance in the end. That’s because the Load balancer we built in this research are establishing connection using Kubernetes’s API, which will return a socket that allow us to connect to the Pod directly. If Load Balancer received tons of requests from clients without limitation and trying to request socket from Kubernetes API. It will cause low performance and stuck the system. Also became the causes of the not ideal response time. In this thesis we’ve evaluated on threads with limitations. To find out the optimized threads number.
    显示于类别:[資訊管理研究所] 博碩士論文

    文件中的档案:

    档案 描述 大小格式浏览次数
    index.html0KbHTML88检视/开启


    在NCUIR中所有的数据项都受到原著作权保护.

    社群 sharing

    ::: Copyright National Central University. | 國立中央大學圖書館版權所有 | 收藏本站 | 設為首頁 | 最佳瀏覽畫面: 1024*768 | 建站日期:8-24-2009 :::
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 隱私權政策聲明