Abstract. In this paper we present an intelligent multi-topology IGP (MT-IGP) based intra-domain traffic engineering (TE) scheme that is able to handle unexpected traffic fluctuations with near-optimal network performance. First of all, the network is dimensioned through offline link weight optimization using Multi-Topology IGPs for achieving maximum path diversity across multiple routing topologies. Based on this optimized MT-IGP configuration, an adaptive traffic engineering algorithm performs dynamic traffic splitting adjustment for balancing the load across multiple routing topologies in reaction to the monitored traffic dynamics. Such an approach is able to efficiently minimize the occurrence of network congestion without the necessity of frequently changing IGP link weights that may cause transient forwarding loops and routing instability. Our experiments based on real network topologies and traffic matrices show that our approach has a high chance of achieving nearoptimal network performance with only a small number of routing topologies.

Abstract. The next generation Internet is designed to accommodate flows that span across multiple domains with quality of service guarantees, in particular bandwidth. In this context, destinations for inter-domain traffic may be reachable through multiple egress routers within a domain. In this paper, we formulate a bandwidth guaranteed egress router selection problem. The objective is to, for each aggregated inter-domain traffic flow, select an egress router that satisfies the end-to-end bandwidth requirement while optirnizing the network resource utilization by which we consider three objective functions: rninirnizing the total bandwidth consumption, improving intra-domain and inter-domain Ioad balancing in the network. We propose a heuristic algorithm with five egress router selection policies to solve this problem. The evaluation of these egress router selection policies through simulation benefits ISPs by choosing the one that fits their target objecti ves.

Abstract. This paper proposes an integrated network management framework for interdomain outbound traffic engineering. The framework consists of three functional blocks (monitoring, optimization and implementation) to make the outbound traffic engineering adaptive to network condition changes such as inter-domain traffic demand variation, routing changes and link failure. The objective is to keep inter-domain link utilization load balanced under any of these changes while reducing service disruptions and reconfiguration overheads. Simulation results demonstrate that using the proposed framework can successfully achieve better load balancing with less service disruptions and re-configuration overheads compared to the alternative approaches. IntroductionOutbound Traffic Engineering (TE) [1,2,3,4] has become increasingly important and been well studied, and is a set of techniques for controlling traffic exiting a domain by assigning the traffic to the best egress points (i.e. routers or links).. The general problem formulation of outbound TE is: given the network topology, BGP routing information and inter-domain Traffic Matrix (TM), determine the best Egress Point (EP) for each traffic demand so as to optimize the overall network performance [2]. Since inter-domain links are the most common bottlenecks in the Internet [2], optimizing their resource utilization becomes a key objective of outbound TE.In practice, network conditions change dynamically, which can make the current outbound TE solutions obsolete and subsequently cause some inter-domain links to become congested over time. One such dynamic change is inter-domain traffic variation, which is typically caused by changes in user or application behavior, adaptations from the TCP congestion control or even routing changes from other domains [5]. In addition to these traffic variations, transient and non-transient inter-domain peering link failures might occur. According to [7] transient inter-domain link failures are common events and their duration is less than a few minutes. Upon failure on a peering link, there may be a large amount of traffic shifted to other available EPs, potentially leading to congestion on these new serving EPs if they are not carefully chosen. In theory, although it is possible to perform outbound TE based on the other proposals in the literature [2,3,4] whenever any of those changes occur, it may require huge computational overheads and a large number of EP reconfigurations given that previous proposals have not considered the reduction of reconfiguration changes and overheads. This can lead to excessive service disruptions and is not practical. As a consequence, lack of TE solutions that react to those dynamic changes rapidly will leave the network unmanaged. It is thus the focus of this paper to make outbound TE more adaptive to fast-changing IP networks by taking into consideration practical network operation and management constraints such as time-efficiency, reconfiguration overheads and service disruptions.In this paper, we pro...

Abstract. The replica server placement problem determines the optimal location where replicated servers should be placed in content distribution networks, in order to optimize network performance. The estimated traffic demand is fundamental input to this problem and its accuracy is essential for the target performance to be achieved. However, deriving accurate traffic demands is far from trivial and uncertainty makes the target performance hard to predict. We argue that it is often inappropriate to optimize the performance for only a particular set of traffic demands that is assumed accurate. In this paper, we propose a scenario-based robust optimization approach to address the replica server placement problem under traffic demand uncertainty. The objective is to minimize the total distribution cost across a variety of traffic demand scenarios while minimizing the performance deviation from the optimal solution. Empirical results demonstrate that robust optimization for replica server placement can achieve good performance under all the traffic demand scenarios while nonrobust approaches perform significantly worse. This approach allows content distribution providers to provision better and predictable quality of service for their customers by reducing the impact of inaccuracy in traffic demand estimation on the replica server placement optimization.

Abstract