臺灣大學: 電信工程學研究所蔡志宏蔡治緯Tsai, Zhi-WeiZhi-WeiTsai2013-03-272018-07-052013-03-272018-07-052011http://ntur.lib.ntu.edu.tw//handle/246246/252808隨著雲端運算科技的快速發展，我們可以發現市場上已經存在了許多商業雲端運算的營運商，例如： Amazon, Google, Azure。他們供應了大規模的運算資源。然而，這些雲端資源的供應商提供了固定價格的資源供使用者存取， 彷彿資源是無限的。有鑑於資源是有限的，由於以競標來競爭有限資源已被視為一有效的資源分配方式，故當雲端資源有限時，以競標機制來分配雲端資源應該是存在需求的。 於此論文中，我們使用了競標機制來為雲端資源動態定價，我們也研究了雲端資源供應商如何在此競標環境中做出最佳的決策，即「保留標」的決定。除此之外，我們使用了模擬的方法在總需求量隨機的環境之下展示此競標機制的結果。我們發現了在競標機制下的雲端資源分配，供應者與使用者行為之間存在了許多有趣的關係。As the Cloud Computing technology and industry evolves rapidly in recent years, one can observe that there already exists many commercialized cloud providers such as Amazon, Google, Azure in the market, offering a large scale of computing resources. However, these providers currently only offer the cloud computing resources by the fixed price approach to users, as if there are unlimited resources. Since the auction approach is already shown to be an efficient approach in many other resource-limited environments, there should be a need for developing such mechanism when the cloud is resource-constrained. In this thesis, we have developed a bid-proportional auction model which brings dynamic price of the computing resource. We also study how to decide the optimal bid for the users and the optimal strategy of releasing computing resource for the providers in the auction-based resource allocation market. Furthermore, we used simulation to illustrate how the model operates in a dynamic and stochastic demand environment. We also discover several interesting relationships between the users and the providers regarding the dynamic price and the total supply-demands situation.6302511 bytesapplication/pdfen-US雲端運算資源分配最佳化拍賣Cloud ComputingResource AllocationOptimizationAuctionthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/252808/1/ntu-100-R97942123-1.pdf