- 學位論文
虛擬化技術下虛擬CPU資源配置之最佳化
Optimize Virtual CPU Resource Allocation in Virtualized Technology
摘要
隨著近年來雲端運算的發展，多核心技術快速進步，使得硬體效能不斷提升，虛擬化成為重要關鍵技術，而如何讓資源得到最佳的配置，成為我們需要探討的議題。 本論文提出建構於KVM(Kernel based Virtual Machine)全虛擬化架構下，虛擬CPU資源配置之最佳化。此系統改善傳統資源配置的缺失，並且針對相關研究DAVMCR機制與DRAMS方案加以修改，採用了CPU SPEC技術解決依照使用者硬體環境的不同來動態調整配置資源， 此外，加入CPU最佳使用率讓整體系統在該設定下達到負載平衡。 實驗結果顯示，此方法可以提高系統資源使用率，比起傳統的配置方案效能提升了34%，達到資源配置最佳化與自動化目標，同時能減少實體主機數量的需求和維修成本、人力支出，充分利用整體系統資源。
並列摘要
With the development of cloud computing, virtualization has become an important key technology. How to solve the resource allocation and utilization problem will be dis-cussed. In this thesis, we presents a dynamic adjustment mechanism for computing re-sources based on KVM (Kernel based Virtual Machine). This system is to improve the shortcomings of the resources allocation in traditional way, and be modified for the study DAVMCR mechanism and the DRAMS program. When system is running, using CPU SPEC technical to solve depending on the user different hardware environment to dy-namic adjust the allocation of resources. In addition, adding the optimal utilization rate of CPU let whole system to achieve load balance. Experimental result shows that by applying the proposed method can get better uti-lization rate of resources. The performance improves 34% compared to the traditional al-location program. It can also reduce the demand of the physical machines and reduce the cost of hardware maintenance. Take full advantage of the overall system resources.