虛擬機器技術讓實體機器可以透過軟體的方式切割成單一獨立的虛擬機器，因此允許多人共享一台實體機器，並擁有各自獨立的計算環境。近來虛擬技術相關研究議題逐漸轉向更高階層的管理方式，也就是以虛擬機房(或虛擬資料中心)為單位進行管理。該虛擬私有機房是一個虛擬混合雲 (Virtual Hybrid Cloud)的環境，讓使用者得以在公開雲的基礎硬體上建立虛擬私有雲，使用者可以透過VPN的技術使用公開雲的資源，且又能在網路層級隔離虛擬私有雲的計算。在這個研究中，我們嘗試使用現有的虛擬化技術來開發一款提供虛擬機房服務平台的建置與管理軟體，我們稱之為SAMEVED ( System Architecture for Managing and Establishing Virtual Elastic Datacenters )。SAMEVED是一個非商業化的系統，而使用者被定位在非營利且有保密需求的單位上，例如學校的網路安全研究。當學校中的老師或學生需要做網路安全研究實驗時，就可以利用SAMEVED，根據自己的實驗需求，簡單且快速地建立一個虛擬的私有機房。在這篇論文裡，我們提出系統的規劃、設計架構，並提供監測的功能，讓虛擬私有機房的服務可以順暢的運作。我們也提供遠端呼叫程序庫給雲端程式使用，因此雲端程式可透過此技術建立虛擬資料中心，並自行定義出資料中心內的網路拓樸等設定。根據我們觀測的結果，SAMEVED目前運作順暢，因此可以證明本研究所提供的功能是穩定且正確的。The virtualization technology can logically divide a physical machine into some independent virtual machines. Hence, people can share the same physical machine and still work on an independent virtual computing environment. Recently, more and more people focus on the issue of managing the virtual computing environment in a higher granule, that is, to administer virtual private datacenters, each of which consists of a collection of virtual machines connected by a virtual local area network. In this research, we use the virtualization technology to develop a software package, SAMEVED (System Architecture for Managing and Establishing Virtual Elastic Datacenters), which can create, deploy, and manage virtual private datacenters. The SAMEVED also provides users the monitoring feature that reports the status of their own virtual private datacenters. The SAMEVED also supports remote procedure calls for authenticated cloud applications, such that the applications can create and modify the configurations of their virtual datacenters. We have launched a cloud eLearning service that aims to support network security experiments on the cloud. The eLearning service requires network isolation in the data link layer. When the teachers or the students try to do a network security experiment such as DDoS, they can create their own virtual private datacenter based on their requirements, without worrying about that the experiment becomes uncontrollable. According to our monitoring data, the eLearning service have created and deployed tens of virtual private datacenters on the SAMEVED. The observed data show that the system is stable and correct.
