【摘要】：自然或者人為的災(zāi)難發(fā)生,使得災(zāi)難生存能力成為了通信網(wǎng)絡(luò)中的主要挑戰(zhàn)。數(shù)據(jù)中心網(wǎng)絡(luò)提供的云服務(wù)為網(wǎng)絡(luò)的生存性帶來了新的機會。現(xiàn)在典型的網(wǎng)絡(luò)結(jié)構(gòu)是基于WDM的數(shù)據(jù)中心網(wǎng)絡(luò),而波分復(fù)用技術(shù)已經(jīng)不能滿足海量的信息需求。已經(jīng)有相關(guān)的研究人員提出彈性光數(shù)據(jù)中心網(wǎng)絡(luò),該網(wǎng)絡(luò)能夠更加精細(xì)的分配頻譜,同時又能支持云服務(wù)。彈性光數(shù)據(jù)中心網(wǎng)絡(luò)的生存性已經(jīng)成為了熱點問題。本文首先介紹彈性光數(shù)據(jù)中心網(wǎng)絡(luò)的特點和研究現(xiàn)狀,介紹了彈性光數(shù)據(jù)中心網(wǎng)絡(luò)的基礎(chǔ)知識和技術(shù),重點討論了彈性光數(shù)據(jù)中心網(wǎng)絡(luò)的路由和頻譜分配等問題。針對容量受限的彈性光數(shù)據(jù)中心網(wǎng)絡(luò),本文基于整數(shù)線性規(guī)劃(Integer Linear Programming,ILP)針對共享保護(hù)和專有保護(hù)提出算法來求解網(wǎng)絡(luò)的最大吞吐量和最小頻譜資源消耗量,并進(jìn)行對比。仿真結(jié)果顯示,共享保護(hù)的ILP算法在相同的網(wǎng)絡(luò)吞吐量情況下,使用更少的頻譜資源。這是因為在共享保護(hù)策略下,備用路徑的鏈路能共享頻譜資源,所以在相同的網(wǎng)絡(luò)吞吐量下,共享保護(hù)所使用的頻譜資源遠(yuǎn)遠(yuǎn)小于專有保護(hù)。為了進(jìn)一步節(jié)省網(wǎng)絡(luò)的頻譜資源,本文在共享保護(hù)網(wǎng)絡(luò)中,基于ILP比較任播和單播兩種保護(hù)方式。由于在任播方式下,工作路徑的目的節(jié)點是兩個數(shù)據(jù)中心任意選擇一個,而單播方式的源節(jié)點和目的節(jié)點都是確定的,在路徑選擇上,任播方式會更加的靈活。仿真結(jié)果顯示基于任播方式的共享保護(hù)ILP算法與基于單播方式的共享保護(hù)ILP算法有著相同的網(wǎng)絡(luò)吞吐量,但是任播方式使用的頻譜資源顯著低于單播方式。
[Abstract]:Natural or man-made disasters make disaster survivability a major challenge in communication networks. The cloud service provided by the data center network brings new opportunities for the survivability of the network. At present, the typical network structure is the data center network based on WDM, and wavelength division multiplexing (WDM) technology can not meet the needs of massive information. Some researchers have proposed a network of flexible optical data centers, which can distribute spectrum more finely and support cloud services. The survivability of elastic optical data center network has become a hot issue. This paper first introduces the characteristics and research status of elastic optical data center network, introduces the basic knowledge and technology of elastic optical data center network, and discusses the routing and spectrum allocation of elastic optical data center network. For elastic optical data center networks with limited capacity, this paper proposes an algorithm based on integer linear programming (Integer Linear Programming,ILP) for shared and proprietary protection to solve the maximum throughput and minimum spectrum resource consumption of the network. Simulation results show that the shared protection ILP algorithm uses less spectrum resources under the same network throughput. This is because the link of the alternate path can share the spectrum resources under the shared protection strategy, so the spectrum resource used by the shared protection is much smaller than that of the proprietary protection under the same network throughput. In order to further save the spectrum resources of the network, this paper compares two kinds of protection schemes based on ILP, anycast and unicast in the shared protection network. In any broadcast mode, the destination node of the working path is chosen arbitrarily by two data centers, while the source node and the destination node of the unicast mode are determined. In the path selection, any broadcast mode will be more flexible. The simulation results show that the shared protection ILP algorithm based on any broadcast mode has the same network throughput as the ILP algorithm based on unicast mode, but the spectrum resource used in any broadcast mode is significantly lower than that in unicast mode.
【學(xué)位授予單位】：深圳大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TN929.1;TP308

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 江南;數(shù)據(jù)中心如何應(yīng)付管理挑戰(zhàn)[J];互聯(lián)網(wǎng)周刊;2001年40期

2 ;簡化管理挑戰(zhàn)——惠普推實用數(shù)據(jù)中心解決方案[J];每周電腦報;2001年67期

3 李慶莉;去數(shù)據(jù)中心看一看——中國銀行華北信息中心計劃處處長云恩善談數(shù)據(jù)中心運行、管理[J];中國金融電腦;2002年12期

4 馬天蔚;;數(shù)據(jù)中心按需造[J];每周電腦報;2002年25期

5 戚麗,蔣東興,武海平,馮珂;校園數(shù)據(jù)中心建設(shè)與管理方法的探索[J];教育信息化;2002年S1期

6 何俊山;您企業(yè)的數(shù)據(jù)中心2003了嗎?[J];微電腦世界;2003年17期

7 ;挖潛數(shù)據(jù)中心[J];金融電子化;2004年07期

8 王琨月;;數(shù)據(jù)中心業(yè)務(wù)就緒[J];每周電腦報;2004年21期

9 包東智;新熱點:創(chuàng)建下一代數(shù)據(jù)中心[J];上海信息化;2005年10期

10 ;把握數(shù)據(jù)中心建設(shè)五大看點[J];中國計算機用戶;2005年10期

相關(guān)會議論文 前10條

1 姚，

本文編號：2209119