一個基於網路行為的雲端服務佈署策略

Abstract

雲端平台上的公用運算近年來成為熱門議題，廣受企業界與學術界所討論，但是一個服務建置在雲端需要花費成本，目前卻未有研究針對雲端服務的成本與風險進行討論，以及規劃佈署的策略。因此本研究主要分析網路行為，分別評估服務佈署在傳統本地端主機與雲端平台上的成本與風險(服務阻塞率)，並進行比較，以規劃合適的佈署策略。針對雲端上的平台即服務架構(Platform as a Service, PaaS)進行探討，我們提出一個佈署策略，此策略利用二個通用的數學模型去分析傳統本地端主機與雲端的成本及顧客流失率。在實驗中運用國立交通大學資訊與財金管理學系與國立交通大學首頁的實際存取記錄數據作為研究對象，評估將此二個網路服務建置於Google 應用服務引擎(Google App Engine, GAE)中所需的成本和可能造成的風險。以國立交通大學為例，在成本方面，推導結果得出若該網站佈署於雲端，可比佈署在傳統本地端主機節省47%以上的支出，另外對於風險也影響極低，顯示該服務是適合建置於雲端上，另外國立交通大學資訊與財金管理學系網站佈署於雲端後也會有低成本與低風險的優點，因此，藉由本研究中建立的策略，可提供給管理者或專家從成本和風險來評估服務是否適合放在雲端的參考。

In recent years, utility computing has become one of the popular topics, and is widely explored in academic and industry. It will be required fees if services deploy in cloud. However, in present researches, there are no discussions about costs and risks of deploying cloud services. In this thesis, we analyze network behavior to assess and compare the costs and risks (Service Blocking Rate, SBR) between traditional local server and cloud, and then can get appropriate deployment strategy. This thesis proposes a deployment strategy which utilizes two mathematical models to analyze costs and risks of services in traditional local server and Platform as a Service (PaaS) platform in cloud. Besides, select two websites as research objectives and analyze the costs and risks of deploying services in Google App Engine (GAE) in simulation. Those websites are: (1) Information and Finance of Management (IFM) in National Chiao Tung University (NCTU) website log files and (2) NCTU website log files. Take NCTU website as an example. If the website deployed in GAE, NCTU will save over 47% of expenditure than deployed in traditional local server, and has low impact in risk. Therefore, NCTU service is obviously appropriate in the cloud. IFM is also suitable for deployed in cloud with low costs and low risk. Based on the strategy, this thesis proposes a suggestion to managers and professionals.