随着互联网的飞速发展,集群结构的下一代核心路由器已经成为研究的重点.在可扩展路由器中(clus- ter router),并行路由算法是关键问题之一.对于广泛部署的OSPF协议,最短路径树(SPT)的并行计算是其并行化的核心难点.本文提出了一种计算...随着互联网的飞速发展,集群结构的下一代核心路由器已经成为研究的重点.在可扩展路由器中(clus- ter router),并行路由算法是关键问题之一.对于广泛部署的OSPF协议,最短路径树(SPT)的并行计算是其并行化的核心难点.本文提出了一种计算最短路径树的算法-分区Dijkstra算法(D-D),分析了算法性能,并通过模拟实验验证了算法的性能.展开更多
Networking plays a crucial role in cloud computing especially in an inter-cloud environment, where data communications among data centers located at different geographical sites form the foundation of inter-cloud fede...Networking plays a crucial role in cloud computing especially in an inter-cloud environment, where data communications among data centers located at different geographical sites form the foundation of inter-cloud federation. Data transmissions required for inter-cloud federation in the complex inter-cloud networking system are often point-to-multi points, which calls for a more effective and efficient multicast routing algorithm in complex networking systems. In this paper, we investigate the multicast routing problem in the inter-cloud context with K constraints where K ≥ 2. Unlike most of existing algorithms that are too complex to be applied in practical scenarios, a novel and fast algorithm for establishing multicast routing tree for interclouds is proposed. The proposed algorithm leverages an entropybased process to aggregate all weights into a comprehensive metric, and then uses it to search a multicast tree(MT) on the basis of the shortest path tree(SPT). We conduct complexity analysis and extensive simulations for the proposed algorithm from the approximation perspective. Both analytical and experimental results demonstrate that the algorithm is more efficient than a representative multi-constrained multicast routing algorithm in terms of both speed and accuracy, and thus we believe that the proposed algorithm is applicable to the inter-cloud environment.展开更多
软件定义网络(Software Defined Network,SDN)以其强大的可编程性和集中控制的优势得到了学术界的广泛关注。现有的SDN设备在执行报文转发时仍然使用最短路径协议,当最短路径中的结点发生故障时,网络仍然需要重新收敛,在此期间报文可能...软件定义网络(Software Defined Network,SDN)以其强大的可编程性和集中控制的优势得到了学术界的广泛关注。现有的SDN设备在执行报文转发时仍然使用最短路径协议,当最短路径中的结点发生故障时,网络仍然需要重新收敛,在此期间报文可能会被丢弃,进而无法传递至目的结点,给实时性应用的流畅性造成了冲击,影响用户体验。学术界普遍采用路由保护的方案来应对网络故障,现有的路由保护方案存在以下两个方面的问题:(1)故障保护率低;(2)当网络出现故障时,备份路径可能会出现路由环路。为了解决上述两个问题,首先提出了备份下一跳计算规则;然后基于此规则设计了一种软件定义网络下的高故障保护率的路由保护算法(Routing Protection Algorithm with High Failure Protection Ratio,RPAHFPR),该算法融合了路径生成算法(Path Generation Algorithm,PGA)、旁支优先算法(Side Branch First Algorithm,SBF)和环路规避算法(Loop Avoidance Algorithm,LAA),可以同时解决已有路由保护方法面临的故障保护率低和路由环路问题;最后在大量的真实网络拓扑和模拟网络拓扑中验证了RPAHFPR方案的性能。与经典的NPC和U-TURN相比,RPAHFPR的故障保护率分别提高了20.85%和11.88%,并且在86.3%的拓扑中可以达到100%的故障保护率,在所有拓扑中可以达到99%以上的故障保护率。RPAHFPR的路径拉伸度基本接近1,不会引入过多的时间延迟。展开更多
文摘随着互联网的飞速发展,集群结构的下一代核心路由器已经成为研究的重点.在可扩展路由器中(clus- ter router),并行路由算法是关键问题之一.对于广泛部署的OSPF协议,最短路径树(SPT)的并行计算是其并行化的核心难点.本文提出了一种计算最短路径树的算法-分区Dijkstra算法(D-D),分析了算法性能,并通过模拟实验验证了算法的性能.
基金supported by the National Natural Science Foundation of China(61309031)
文摘Networking plays a crucial role in cloud computing especially in an inter-cloud environment, where data communications among data centers located at different geographical sites form the foundation of inter-cloud federation. Data transmissions required for inter-cloud federation in the complex inter-cloud networking system are often point-to-multi points, which calls for a more effective and efficient multicast routing algorithm in complex networking systems. In this paper, we investigate the multicast routing problem in the inter-cloud context with K constraints where K ≥ 2. Unlike most of existing algorithms that are too complex to be applied in practical scenarios, a novel and fast algorithm for establishing multicast routing tree for interclouds is proposed. The proposed algorithm leverages an entropybased process to aggregate all weights into a comprehensive metric, and then uses it to search a multicast tree(MT) on the basis of the shortest path tree(SPT). We conduct complexity analysis and extensive simulations for the proposed algorithm from the approximation perspective. Both analytical and experimental results demonstrate that the algorithm is more efficient than a representative multi-constrained multicast routing algorithm in terms of both speed and accuracy, and thus we believe that the proposed algorithm is applicable to the inter-cloud environment.
文摘软件定义网络(Software Defined Network,SDN)以其强大的可编程性和集中控制的优势得到了学术界的广泛关注。现有的SDN设备在执行报文转发时仍然使用最短路径协议,当最短路径中的结点发生故障时,网络仍然需要重新收敛,在此期间报文可能会被丢弃,进而无法传递至目的结点,给实时性应用的流畅性造成了冲击,影响用户体验。学术界普遍采用路由保护的方案来应对网络故障,现有的路由保护方案存在以下两个方面的问题:(1)故障保护率低;(2)当网络出现故障时,备份路径可能会出现路由环路。为了解决上述两个问题,首先提出了备份下一跳计算规则;然后基于此规则设计了一种软件定义网络下的高故障保护率的路由保护算法(Routing Protection Algorithm with High Failure Protection Ratio,RPAHFPR),该算法融合了路径生成算法(Path Generation Algorithm,PGA)、旁支优先算法(Side Branch First Algorithm,SBF)和环路规避算法(Loop Avoidance Algorithm,LAA),可以同时解决已有路由保护方法面临的故障保护率低和路由环路问题;最后在大量的真实网络拓扑和模拟网络拓扑中验证了RPAHFPR方案的性能。与经典的NPC和U-TURN相比,RPAHFPR的故障保护率分别提高了20.85%和11.88%,并且在86.3%的拓扑中可以达到100%的故障保护率,在所有拓扑中可以达到99%以上的故障保护率。RPAHFPR的路径拉伸度基本接近1,不会引入过多的时间延迟。
