Hadoop是一個open source的平台，可以用以執行MapReduce的平行分散式程式以處理大量資料及進行平行化運算。本研究希望能利用Hadoop來實現平行化雲端TaaS測試平台，期望能達成TaaS-based測試工具，並用來進行效能及壓力測試(stress test) 。我們運用Hadoop cluster的computation power來模擬大量使用者連線、大量使用者上傳或下載、大量的Web services同時呼叫等。並針對各式測試模式訂定測試腳本(testing script)標準，使Client可以描述欲進行的network transactions，使用者透過使用介面將測試腳本傳入系統，啟動後台Hadoop cluster進行平行測試。使用者（Client）不需要知道如何撰寫Hadoop MapReduce程式就可以進行各式平行測試。首先使用者先將預測的細節透過我們實作的網頁介面寫成測試腳本，經過一連串的腳本分析，確定無誤後，將testing script傳送給testing master以啟動此testing service。Testing master根據testing script的需求啟動Hadoop cluster中的Hadoop nodes將遠端呼叫送給被測系統，並收集測試結果，然後將統計的結果傳回給使用者。

Hadoop is an open source platform, and it can use to perform MapReduce program by its powerful concurrent computation to deal with big data. Our research is performing TaaS-based testing system by Hadoop, and doing stress testing and performance testing. We use computation power of Hadoop cluster to simulate lots of users that call amount of network service like Web service, upload , download, and network connections at the same time. And We also define the standard of the testing script to describe these network transactions. Clients do not know how to coding a Hadoop MapReduce program, they also can do the stress testing on our system. First, clients can write the testing script that describe the network transactions. After checking the testing script is valid by system, clients can send the testing script to testing master to start the testing service. Testing master start the Hadoop nodes of Hadoop cluster to do the remote call according to testing script, then collecting the response from the remote system. Finally, our system will calculate the response and write a report to the clients.

[1]. L. Yu, W. Tsai, X. Chen, L. Liu, Y. Zhao, L. Tang, and W. Zhao, "Testing as a Service over Cloud," in 2010 Fifth IEEE International Symposium on Service Oriented System Engineering, 2010, pp. 181-188. 
[2]. “Apache’s Jmeter” , http://jmeter.apache.org/
[3]. Hadoop, http://hadoop.apache.org/
[4]. Stress testing , http://en.wikipedia.org/wiki/Stress_testing
[5]. MapReduce技術, http://en.wikipedia.org/wiki/MapReduce
[6]. Google Big Table, http://en.wikipedia.org/wiki/BigTable
[7]. LoadStorm, http://loadstorm.com/
[8]. SOASTA, http://www.soasta.com/
[9]. LoadRunner, http://en.wikipedia.org/wiki/HP_LoadRunner
[10]. Jing, You, et al. "JMeter-based aging simulation of computing system."Computer, Mechatronics, Control and Electronic Engineering (CMCE), 2010 International Conference on. Vol. 5. IEEE, 2010.