由於高速的網際網路發展及普遍化與硬體效能和行動裝置的快速運算需求提升下，網際網路成為未來資訊產業的主流，而雲端運算的想法也因此順應而生，近年來更是各大企業與學術界關注的焦點。在一個傳統的雲端資料中心，所有的資料都存儲在一個集中式的檔案系統，如存儲區域網絡（SAN）檔案系統，並在許多電腦集群(computer clusters)組成一個大的計算池執行計算任務。每當使用者請求計算資源，從集中式檔案系統獲得所需的虛擬機圖像（VM image）,然後在計算池裡初始化虛擬機。當請求量多的時候，在此過程中可能產生效能瓶頸，導致整個資料中心的效能降低，因此使用 Cache 的方法就變得很重要。在這篇論文中，我們提出一些以統計資料或實驗參數來訂定的 Cache 的解決策略，並設計了不同的虛擬機請求的情境，利用cloudsim 雲模擬器去評估在不同情況下不同的 Cache 策略的效能如何。最後找到一個最佳的 Cache 策略與容量，來達到減緩效能瓶頸與提升整個資料中心的效能，並節省企業在 Cache 容量的花費成本。The Internet has become the foundation of the information technology industry, and realized the cloud computing paradigm that provides computing power to users in a pay-as-you-go manner. In a typical cloud data center, all data are stored in a centralized file system such as a storage-area-network (SAN) file system, and computation tasks are executed in a large computing pool consisting of many computer clusters. Whenever a user requests for a computation resource, the management system of the data center should obtain the desired virtual machine (VM) image from the centralized file system and then instantiate the VM on the computing pool for the user. This procedure results in a possible performance bottleneck, and may cause performance degradation of the whole data center. In this paper, we have investigated several caching strategies to see if they can alleviate the data transfer bottleneck between the centralized file system and the computing pool in a data center. We have designed different VM requests scenarios, and used a cloud simulator, cloudsim, to evaluate the performance of different caching strategies under different scenarios. Finally, we pointed out the better caching strategy that can be incorporated in the design of a large data center.
