Abstract — This paper studies load balancing for multipath Internet routing. We focus on hash-based load balancing algorithms that work on the flow level to avoid packet reordering which is detrimental for the throughput of transport layer protocols like TCP. We propose a classification of hash-based load balancing algorithms, review existing ones and suggest new ones. Dynamic algorithms can actively react to load imbalances which causes route changes for some flows and thereby again packet reordering. Therefore, we investigate the load balancing accuracy and flow reassignment rate of load balancing algorithms. Our exhaustive simulation experiments show that these performance measures depend significantly on the traffic properties and on the algorithms themselves. As a consequence, our results should be taken into account for the application of load balancing in practice. I.

Abstract. In this paper we propose a resilient scheme for multi-path routing using a biologically-inspired attractor selection method. The main advantage of this approach is that it is highly noise-tolerant and capable of operating in a very robust manner under changing environment conditions. We will apply an enhanced attractor selection model to multi-path routing in overlay networks and discuss some general properties of this approach based on numerical simulations. Furthermore, our proposal considers randomization in the path selection which reduces the selfishness and improves the overall network-wide performance. 1

Abstract—Flow-based load balancing algorithms for multipath Internet routing are often used for traffic engineering. However, the target load distribution and the load balanced result agree only on average, and there is a significant inaccuracy over time due to stochastic effects. Dynamic load balancing reduces this inaccuracy by relocating flows to other paths in regular time intervals. This causes packet reordering. Therefore, the flow reassignment rate should be kept low. In this paper we consider load balancing in networks. It differs from load balancing at a single node by the fact that several load balancing steps may be performed at consecutive nodes in series. This affects the flow reassignment rate and the load balancing accuracy due to interdependencies and polarization effects. We quantify the impact by simulation results, explain the observed phenomena, and give recommendations for load balancing in practice.

Abstract. IP networks have established as a global telecommunication platform with increasing user population and an extending spectrum of services. The traffic is also steadily increasing, recently driven by peer to peer networking in addition to client server based applications. Network planers and operators have to ensure the scalability of IP platforms in a permanent upgrade process for transmission capacities. At present, Deutsche Telekom and other telecommunication network providers are introducing traffic engineering methods to achieve an optimum resource utilization. In a first step, traffic engineering can be applied to a predefined network topology, but a comprehensive approach has to be coordinated with a process for upgrading the link capacities and has to prepare for relevant failure scenarios. We have evaluated the efficiency of traffic engineering together with simple link upgrade strategies in order to get a maximum throughput. Therefore a predefined traffic matrix T is taken into account, which is scaled by a maximum factor λmax such that the traffic demand λmax T can still be carried on the available network resources. The influence of the network topology on the evaluation results is shown in examples with regard to single link failures.

Abstract — Multipath structures are the base for many recently developed rerouting and protection switching mechanisms. All of these methods show a similar path layout, rely on traffic distribution, and promise resilience with only little backup capacity. Therefore, it is hard to recognize their commonalities and differences at first sight. This paper provides an overview of these related mechanisms and a comparative analysis regarding their applicability in optical and packet switched technologies, their path layout, their reaction time, their dynamic adaptability, and many other aspects. I.

Abstract—The self-protecting multipath (SPM) is a simple protection switching mechanism that can be implemented, e.g., by MPLS. We present a linear program for the optimization of the SPM load balancing parameters to maximize the amount of transportable traffic with resilience requirements. This is needed to configure the SPM for the deployment in legacy networks. Our study shows that the SPM is very efficient in the sense that it can carry 50 %- 200 % more protected traffic than IP rerouting in sufficiently meshed networks. The investigation of the computation time and the memory consumption recommends the COIN LP (CLP) as preferred LP solver. The computation time of the program depends mainly on the number of links in the network and networks with up to 240 links can be optimized within one hour on a standard PC. I.

Abstract — The self-protection multipath (SPM) is a simple protection switching mechanism. It distributes the traffic over several disjoint paths from source to destination according to a traffic distribution function. When a path fails, the traffic is redistributed to the working paths according to another traffic distribution function, i.e., the traffic distribution function depends on the failed path. The contribution of this work is the introduction of a failure-specific traffic distribution function for the SPM that depends on the exact failure of the paths. We present a linear program for the global optimization of the traffic distribution function of all SPMs in all protected failure scenarios. Finally, we compare the amount of protected traffic that can be transported in the network for the conventional SPM and the new failure-specific SPM (FSPM). I.

Abstract. The self-protecting multipath (SPM) is a simple and efficient endto-end protection switching mechanism for transport networks. It distributes the traffic of a demand between two nodes according to a specific load balancing function over several disjoint paths and redistributes it if one of them fails. The load balancing functions can be optimized so that backup capacity in the network is optimally shared by multiple demands in various failure scenarios. As a result, resilience against all link and node failures can be achieved with only little extra capacity and in capacitated networks more protected traffic can be carried with the SPM than with other resilience mechanisms. The SPM is rather simple which facilitates its deployment in practice. This chapter explains the SPM in detail, distinguishes it from other, similar mechanisms, shows how the load balancing functions can be optimized, and illustrates the superior performance of the SPM. 1 Structure and Operation The self-protecting multipath (SPM) has been first published in [1]. It carries a traffic demand d between two routers in a network and protects the transmission against

In this paper, we consider adaptive bandwidth allocation (ABA) for capacity tunnels as an effective means for multi-hour network design. Traffic engineering (TE) tunnels established in a network from border-to-border (b2b) can be used not only for route pinning between ingress/egress node pairs but also for efficient implementation of resilient network admission control. If static bandwidth allocation (SBA) based on peak-rate traffic assumptions is used to dimension the b2b tunnels, fluctuations of the network traffic can lead to under- or overprovisioning of network capacity in the tunnels. If ABA is used instead, the tunnel sizes are dynamically adapted to current traffic conditions. The efficient use of network capacity assigned to TE tunnels strongly depends on the structure of these tunnels. The contribution of this paper is an assessment of the bandwidth savings that are achievable with ABA in comparison to SBA for various tunnel structures with different path layouts and load balancing strategies. Our results show that the capacity savings due to ABA depend on the routing and load balancing schemes provisioned in the network and that these savings may be increased by appropriately chosen tunnel implementations. 1

In this work, we present several end-to-end protection switching mechanisms for application in Multiprotocol Label Switching (MPLS). In case of local outages in the network, they deviate the traffic around the failed element over backup paths. They are easy to implement and reduce the additional capacity to maintain the Quality of Service (QoS) on the backup paths. We study the capacity savings of the presented methods for various protection schemes with different traffic matrices. We further test the influence of different resilience constraints such as the set of protected failure scenarios, bandwidth reuse restrictions due to optical communication, and traffic reduction due to failed border routers.