透過您的圖書館登入
IP:3.140.185.170
  • 學位論文

在雲端中具可存活性之虛擬資料中心資源配置

Resource Allocation for Survivable Virtual Data Center in Clouds

指導教授 : 廖婉君

摘要


近年來,隨著雲端資料中心開始成為服務供應商佈署各種多樣化服務的熱門選擇,為了能夠有效地代表各種應用服務對於雲端資源的需求,虛擬資料中心的概念被提出來以協助雲端供應商針對不同服務分配雲端資源;另一方面,以資料中心的規模來看,數以萬計的伺服器、交換器以及實體線路都有可能發生故障,這些硬體設備的故障是相當頻繁而且無可避免的,連帶使得佈署在雲端上的應用服務效能下降,甚至造成服務的中斷。因此,在這篇論文中,我們提出了在雲端上具可存活性之虛擬資料中心資源配置的問題,目的是在保證軟體供應商所要求的虛擬資料中心在任意的硬體設備(包含伺服器與交換器)故障之下,仍然能滿足其應用服務之需求,且為了讓雲端供應商能夠容納更多虛擬資料中心,此問題的目標追求讓整體資料中心的網路頻寬使用量最小化。透過對問題的分析,我們證明了此問題為一多項式困難之問題,並推導了不同交換器故障對於資料中心網路的效能影響。接著,我們提出了一個兩階段演算法來配置虛擬資料中心,第一階段為虛擬機器配置,第二階段為虛擬流量映射,針對第一階段的虛擬機器配置,我們提出了多項式時間的演算法來減輕網路的負擔;至於第二階段的虛擬流量映射,在給定一組虛擬機器配置之下,我們建立了一個線性規劃的模型來考慮預約頻寬的分享,進而達成最小化頻寬使用量的目的。透過模擬分析,平均來說,我們的演算法比起基準演算法可以節省約43% 的頻寬消耗;即使與不具可存活性的虛擬資料中心作比較,我們僅需要額外的12% 網路資源,就可以保證在任意的硬體故障下,我們的資源配置方案依然能夠滿足虛擬資料中心所要求的服務品質。

並列摘要


Data centers have become a popular infrastructure to host diversified application services from tenants. To provide resource guarantee, virtual data center (VDC) is proposed to allocate virtual machines (VM) and network bandwidth for tenants. On the other hand, at cloud scale, hardware failures are common and inevitable which lead to degradation in service performance. To address this challenge, we propose survivable virtual data center allocation problem (SVDCAP) in this thesis, which aims at allocating survivable virtual data center (SVDC) to guarantee that resource demands are always satisfied before and after failures. Our objective is to minimize total bandwidth consumption in order to accept more SVDCs. We prove the NP-hardness for the problem and analyze the failure impacts on bandwidth loss to examine feasibility. Our algorithm solves the problem in two stages: VM placement (VMP) and virtual link mapping (VLM). For VMP, considering VM collocation, polynomial-time algorithms are designed to reduce network load. In addition, a linear programming model is established to efficiently reserve bandwidth in VLM stage. Extensive simulation results demonstrate that our algorithm saves 43% network resource compared to the baseline algorithm under different loads. To guarantee survivability, we use only 12% more network resource compared with regular VDC communication.

參考文獻


[1] M. F. Bari et al., "Data Center Network Virtualization: A Survey," IEEE Communications Surveys Tutorials, vol. 15, no. 2, pp. 909-928, Second Quarter 2013
[2] A. Greenberg et al., "VL2: A Scalable and Flexible Data Center Network," in Proc. ACM SIGCOMM’09, Barcelona, Spain, Aug. 2009.
[3] C. Guo, et al., "BCube: A High Performance, Server-Centric Network Architecture for Modular Data Centers," in Proc. ACM SIGCOMM’09, Barcelona, Spain, Aug. 2009.
[4] M. Al-Fares, A. Loukissas, and A. Vahdat, "A Scalable, Commodity Data Center Network Architecture," in Proc. ACM SIGCOMM’08, Seattle, USA, Aug. 2008.
[5] C. Guo et al., "SecondNet: a Data Center Network Virtualization Architecture with Bandwidth Guarantees," in Proc. ACM International Conference on Emerging Networking Experiments and Technologies (Co-NEXT’10), Philadelphia, USA, Dec. 2010.

延伸閱讀