基於群眾外包的分散式自動化網路服務測試平台

Abstract

在現代生活中，網路科技的發展與普及下，網路服務扮演相當重要的角色，人們在日常中透過各種連網裝置使用網路服務。然而，網路服務在開發過程中，規模會變大，且愈趨複雜，如此一來，待開發完成上線後，使用者可能會因網站內部隱藏的問題，而無法取得服務，導致使用者經驗下滑。為了解決此議題，許多與網路服務測試的研究應用相繼被提出，有部分工具使用單機模式環境進行測試程序，部分運用雲端運算技術，將測試服務部署在虛擬機器上。然而，目前的應用大部分僅放在網站伺服器本身的效能測試上，對於內部問題偵測尚未琢磨太多，再者，測試形式大多不符真實使用情境。因此，提供由內而外且擬真的網路服務測試是必要的。 本論文提出一個基於群眾外包模式的自動化分散測試服務，利用Web介面尋求來自世界各地的網際網路使用者，加入本測試平台成為測試節點或測試者，利用測試節點的運算資源執行測試任務，測試者根據自身需求提出測試要求。目的在於對測試目標網域下的所有資源提出存取要求，自動化地對目標網站伺服器執行非功能性測試，不但能找出網站內部的問題，且了解其效能表現。另外，本論文提出一個模擬真實使用者瀏覽網站的方式，使測試程序符合真實使用情境。透過來自各個地區的效能評估數據，地域性差異測試也能完全得知。至於測試結果，我們提供多樣且詳盡的評測報告，讓測試者完整掌握狀況。

In modern society, web service plays an important role and changes the ways that human beings live in. Thanks to the evolution of electric products and the advance of web technology, people are able to use a bunch of web services by using different types of internet device such as personal computer, notebook, smart TV and smart phone and etc. A lot of things can be done on the internet, which makes it convenient. However, as a web-site becomes larger, users are probably not able to access the service due to some hidden problems. To solve this issue, researches on web service testing are carried out one after another. Some of them perform single-node test. Some use cloud computing technology to build the test environment. However, those tools mostly focus on the performance of web service itself and do not get close to the real condition. Therefore, it is essential to give a comprehensive web service testing which can not only go deep into the core structure to discover potential mistakes but also perform humanized test. In the present thesis, we propose a testing system based on crowdsourcing model to carry out distributed non-functional test on target web server automatically. The system provides an interface for users around the world to join as test nodes or testers. Test nodes contribute their computing resources to help execute test tasks, and testers can submit test requests by configuring the test parameters according to their requirement. By evaluating every single resource of web service automatically, we can not only find out internal problems but also understand the performance of the web service. In order to make the entire test process the same as the real condition, we design a testing scenario which simulates user activity. Additionally, complete geographical test is available with the performance measurements coming from different regions. For test results, we represent detailed reports to give full assessment.