Enhance Stress Testing Power by Synchronizing JMeter Test Scripts

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：43

、訪客IP：3.14.246.254

姓名

李宛娟(Wan-Chuan Lee) 查詢紙本館藏

畢業系所

軟體工程研究所

論文名稱

(Enhance Stress Testing Power by Synchronizing JMeter Test Scripts)

相關論文

★ CRUnit - Capture / Replay Based Unit Testing	★ Locating Interested Code by Program Execution Paths with Debugger
★ An OpenStack Based Testing as a Service Platform	★ Visualize Ripple Effect with Analyzing Object-Oriented Design Relationship
★ Change History Tracing Tool for Arbitrary Programming Language	★ Virtual Objects for Program Visualization in xDIVA
★ 以 GDB 實作 XThreadDebugger-Linux(XTD-Linux)提供 Linux 平台之多執行緒除錯	★ 支援版本控制系統之文件撰寫工具
★ Korat: An O.S.-independent Capture/Replay Test Automation System	★ GUI Component Detection for Cross-Platform Applications–Using Input Device and Image Change Synergistic Detection Method
★ 應用設計模式於 RPA 軟體實作低維護成本的屬性面板自動修正功能	★ 使用靜態分析偵測 JavaScript 應用程式中的 Race Condition
★ 基於物件導向與 Clean Code 概念進行 xDiva 重構與優化	★ 基於xDIVA之利用關鍵影格將3D物件動畫化與即時保存的視覺化工具
★ CoolPCB：以控制點為主的電路板切割成形繪製自動化方法	★ 利用軟體 UI 實現擴充功能突破原始碼限制的工具

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

[檢視]

[下載]

本電子論文使用權限為同意立即開放。
已達開放權限電子全文僅授權使用者為學術研究之目的，進行個人非營利性質之檢索、閱讀、列印。
請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。

摘要(中)

在軟體開發過程中，軟體測試是非常重要的一環，可以確保軟體品質，以及系統的穩定性。為了確保系統可以應付龐大流量，開發團隊通常會運用壓力測試來驗證系統在超出預期負荷的壓力時，系統運作是否正常。
過去開發團隊或測試團隊在進行壓力測試時，往往需要付出非常龐大的運算資源，才能達到一定的同時上線使用者數量來對待測系統進行測試。另外壓力測試常常被過度簡化成使用少數一兩個測試情境來模擬大量使用者，但是一般網頁服務的真實情況則是每個使用者進入網頁服務時，使用者對系統操作的情境並不相同，彼此可能會互相影響，如果能真的如此才能真正的暴露出系統效能瓶頸。
本論文提出一個嶄新的壓力測試技術，這個技術可以同步壓力測試腳本。運用同步壓力測試技術，讓測試人員在進行測試時，可以有效地降低所需要的運算資源，使用較少的運算資源就能夠展現出系統的效能瓶頸。本研究是建立在本實驗室過去開發的TaaS平台[1]上。使用我們系統的測試人員，只需選取過去上傳的一般壓力測試腳本，就能夠輕鬆地對壓力測試腳本進行同步設定，並輕鬆的運用OpenStack自動化的張羅與配置相關硬體準備與設定等問題。

摘要(英)

Software testing is an important part in the software development process. It can ensure the software quality and system stability. In order to ensure that the system can handle huge network traffic, the development team can use stress testing to verify that the system is running normally when incoming traffic exceeds the expected load.
Stress testing is an expensive technique. Emulating a large number of concurrent users requires a lot of computing resources. Besides, traditional stress testing often uses a few test scenarios to simulate a large number of users, but real situation is that each user’s operation is different in a general web service. If we can emulate the interaction of different user scenarios as real as possible, the performance bottlenecks are likely to be exposed.
In this thesis, we propose a brand-new software testing technique. This technique can synchronize the events in different test scripts. When tester or development team use the software testing technique it can reduce computing resource for a stress testing. We hope to find out performance bottleneck of a software system under test with less computing resource. The technique is implemented inside a TaaS web portal developed in our lab. Tester can simply use the synchronizing technique with the help of OpenStack to easily deploying a stress testing.

關鍵字(中)

★ 同步
★ 壓力測試
★ 效能瓶頸

關鍵字(英)

論文目次

目錄
摘要 i
第1章、緒論 1
1-1研究背景與動機 1
1-2論文目標 2
1-3論文架構 3
第2章、研究背景與相關研究 4
2-1軟體測試 4
2-1-1效能測試 4
2-1-2負載測試 5
2-1-3壓力測試 5
2-2壓力測試工具 5
2-2-1 JMeter 5
2-3 OpenStack 7
2-4 Readers-writers lock 8
2-5 網際網路應用程式(Web application)效能問題及瓶頸 9
2-6 MySQL Lock and Transaction 10
第3章、暴露效能瓶頸與壓力測試的運算資源 12
3-1壓力測試需要大量運算資源 12
3-2系統效能瓶頸發生原因 12
3-3壓力測試與效能瓶頸案例 13
案例一、Readers-writer lock問題 13
案例二、過去壓力測試腳本問題 15
3-4壓力測試工具限制 16
3-5同步壓力測試目標 17
第4章、系統功能與介面 19
4-1 TaaS Web Portal介紹 19
4-2同步壓力測試功能與介面設計 24
第5章、同步壓力測試架構與實作 29
5-1 同步壓力測試系統架構 29
5-1-1同步壓力測試之分散式測試架構 30
5-2同步壓力測試腳本設計與實作 31
5-3同步壓力測試腳本分配 33
第6章、同步壓力測試評估 35
6-1實驗環境配置 35
6-2實驗方法與設計 37
6-2-1實驗流程 37
6-2-2測試腳本設計 39
6-3實驗結果 41
6-3-1 OneTicketWritedAtOnce(Normalize) 41
6-3-2 MultiTicketsWritedAtOnce(Larger Lock Range) 43
6-3-3 MultiTicketsWritedAtOnce 45
6-3-4 MultiTicketsWritedAtOnce(LoadBalance) 47
6-4實驗總結 49
第7章、結論與未來展望 50
第8章、參考文獻 51

參考文獻

1] Yeh, S.-W., "An OpenStack Based Testing as a Service Platform", in 軟體工程研究所. 2016, 國立中央大學.
[2] Gheorghiu, G. Performance vs. load vs. stress testing. 2005; Available from: http://agiletesting.blogspot.tw/2005/02/performance-vs-load-vs-stress-testing.html.
[3] Khan, R. and M. Amjad. "Web application′s performance testing using HP LoadRunner and CA Wily introscope tools". in 2016 International Conference on Computing, Communication and Automation (ICCCA). 2016.
[4] Cai, J. and Q. Hu. "Analysis for cloud testing of web application". in The 2014 2nd International Conference on Systems and Informatics (ICSAI 2014). 2014.
[5] Apache JMeter™. Available from: http://jmeter.apache.org/.
[6] LoadRunner. Available from: https://saas.hpe.com/en-us/software/loadrunner.
[7] LoadComplete. Available from: https://smartbear.com/product/loadcomplete/overview/.
[8] What is BeanShell? ; Available from: http://www.beanshell.org/intro.html.
[9] OpenStack. Available from: https://www.openstack.org/.
[10] Software-OpenStack Open Source Cloud Computing Software.
[11] Readers–writer lock. Available from: https://en.wikipedia.org/wiki/Readers%E2%80%93writer_lock.
[12] Web applications Performance Symptoms and Bottlenecks Identification. Available from: http://www.agileload.com/agileload/blog/2012/11/27/web-applications-performance-symptoms-and-bottlenecks-identification.
[13] MySQL InnoDB Storage Engine. Available from: https://dev.mysql.com/doc/refman/5.7/en/innodb-storage-engine.html.
[14] ACID. Available from: https://zh.wikipedia.org/wiki/ACID.
[15] 【直擊Modern Web 2015】解決20萬人秒殺搶票系統架構大揭露. Available from: http://www.ithome.com.tw/news/96113/.
[16] How to scale a Nodejs app based on number of users. Available from: http://adrianmejia.com/blog/2016/03/23/how-to-scale-a-nodejs-app-based-on-number-of-users/

指導教授

鄭永斌

審核日期

2017-7-13

推文