請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/49210
標題: | 次世代無線技術之資料中心網路效能優化 Network Performance Enhancement for Next Generation Wireless Data Centers |
作者: | Ching-Chih Chuang 莊清智 |
指導教授: | 逄愛君 |
關鍵字: | 群播,資料冗餘,無線資料中心,軟體定義網路,SDN延展性, Multicast,Data redundancy,Wireless data centers,Software-defined networking,SDN scalability, |
出版年 : | 2016 |
學位: | 博士 |
摘要: | 隨著雲端服務爆炸性的成長,巨型資料中心廣泛的被建置來提供重要的運算資源並獲得顯著的經濟效益。不幸地,先前的研究報告指出,資料中心網路於資料層與控制層皆面臨到嚴重的網路效能崩塌問題。針對資料層部分,巨量的群播流量造成額外不必要的資料傳輸與網路壅塞的問題。而針對控制層部分,於軟體定義網路(Software-defined networking, SDN)中,當SDN採用三元內容定址記憶體來儲存規則時,單一強大SDN控制器其無法負荷每單位時間大筆資料流所帶來之需求量。有鑑於以上的觀察,我們利用60 GHz無線傳輸技術來解決資料中心的兩個網路議題。為降低整體的群播流量,我們分別制定群播樹建構與維護等問題並同時考量有線與無線連結於無線資料中心網路,同時我們證明此兩個問題皆為NP-hard。為解決群播樹建構與維護的問題,我們提出三個高效能啟發式演算法來重新調整與建構群播樹。而針對控制層問題,我們企圖避免傳送巨量的資料流需求傳送至SDN控制器,以達到提升SDN控制器的延展性為目標。為解決此延展性問題,我們研究最小化資料流需求問題,其主要目標為最小化整體的資料流需求量。接著,我們提出一個同時考量路由與轉送機制來解決此問題。最後,透過一連串的實驗評估來證明我們所提出的方法均能有效解決資料層與控制層所面臨的網路問題。 With the explosive growth of cloud-based services, large- scale data centers are widely built for housing critical computing resources to gain significant economic benefits. Unfortunately, previous reports were shown that the performance of a network has significantly degraded on data and control planes in data center networks. For the data plane, massive multicast traffic may result in unnecessary data transmissions and network congestions. For the control plane, in a software-defined networking, a powerful SDN controller cannot accommodate enormous flow setup requests when the SDN switches use ternary content addressable memory (TCAM) for storing rules. Above reports motivate us to leverage 60 GHz wireless technology to address the two raised issues in data center networks. To reduce the total multicast traffic, we formulate the multicast tree building and maintenance problems with the consideration of coexisting wired and wireless links in wireless data center networks. We prove that the target problems are NP-hard. To tackle the multicast tree building and maintenance problems, we present three efficient heuristic algorithms to reconstruct the multicast trees. As for the control plane, we aim to improve the scalability of SDN controllers, and attempt to avoid sending massive flow setup requests to the controller. To address the scalability issue, we investigate the flow setup minimization problem with the objective of minimizing the total number of flow setup requests. Then, we propose a mechanism with consideration of the routing and forwarding to tackle the problem. A series of simulation results shows that our proposed algorithms are very effective to solve the network problems on data and control planes. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/49210 |
DOI: | 10.6342/NTU201602968 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-105-1.pdf 目前未授權公開取用 | 3.6 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。