Segment Routed Traffic Engineering with Bounded Stretch in Software-Defined Networks
Segment Routed Traffic Engineering with Bounded Stretch in Software-Defined Networks
复制标题
DOI：
10.1109/lcn.2018.8638085
复制
发表时间：
2018-10
期刊：
影响因子：
--
通讯作者：
Tossaphol Settawatcharawanit;Vorapong Suppakitpaisarn;S. Yamada;Yusheng Ji
中科院分区：
文献类型：
--
作者：
Tossaphol Settawatcharawanit;Vorapong Suppakitpaisarn;S. Yamada;Yusheng Ji
分段路由流量工程正在成为网络运营商通过使用分段路由路径作为路由选择的候选来管理资源利用的重要应用。为了促进网络运营商的需求，流量工程方案应该是快速和有效的。这两个特征是非常重要的，因为流量工程程序必须在短时间间隔内周期性地调用。段路由路径可以通过连接两个节点之间的最短路径来构造，使得存在从源到目的地的路径。我们感兴趣的问题，找到中间节点，以最大限度地减少最大的链路利用率，以构建分段路由路径。然而，现有的方法有缺点，要么他们需要大量的时间来找到一个解决方案，或者他们必须牺牲大量的链路利用率。为了解决这些问题，我们建议通过使用相对于源-目的地对的最短路径的有界拉伸约束来限制中间节点候选者的数量。然后，我们根据现有工作评估计算时间和链路利用率。我们表明，有界拉伸约束有助于减少计算时间，同时可以实现接近最佳的链路利用率。
Segment Routed Traffic Engineering is emerging as an important application for network operators to manage resource utilization by using segment routing paths as candidates for route selection. In order to facilitate the network operator demands, a traffic engineering program should be fast and efficient. These two characteristics are very essential since the traffic engineering program must be invoked periodically in short intervals. The segment routing paths can be constructed by concatenating the shortest paths between two nodes such that there is a path from source to destination. We are interested in the problem to find intermediate nodes to construct segment routing paths minimizing the maximum link utilization. However, the existing approaches have the shortcomings that either they require a substantial amount of time to find a solution or they must sacrifice a considerable amount of link utilization. To address these issues, we propose to limit the number of intermediate node candidates by using a bounded stretch constraint relative to the shortest path of the source-destination pair. Then, we evaluate the computation time and link utilization against the existing work. We show that the bounded stretch constraint helps reduce the computation time while a near optimal link utilization can be achieved.