Anders Gunnar

Traffic matrix estimation on a large IP backbone: a comparison on real data

This paper considers the problem of estimating the point-to-point traffic matrix in an operational IP backbone. Contrary to previous studies, that have used a partial traffic matrix or demands estimated from aggregated Netflow traces, we use a unique data set of complete traffic matrices from a global IP network measured over five-minute intervals. This allows us to do an accurate data analysis on the time-scale of typical link-load measurements and enables us to make a balanced evaluation of different traffic matrix estimation techniques. We describe the data collection infrastructure, present spatial and temporal demand distributions, investigate the stability of fan-out factors, and analyze the mean-variance relationships between demands. We perform a critical evaluation of existing and novel methods for traffic matrix estimation, including recursive fanout estimation, worst-case bounds, regularized estimation techniques, and methods that rely on mean variance relationships. We discuss the weaknesses and strengths of the various methods, and highlight differences in the results for the European and American subnetworks.

Data-driven traffic engineering: techniques, experiences and challenges

This paper presents a global view of measurement-driven traffic engineering, explores the interplay between traffic matrix estimation and routing optimization and demonstrates how demand uncertainties can be accounted for in the optimization step to guarantee a robust and reliable result. Based on a unique data set of complete measured traffic matrices, we quantify the demand uncertainties in an operational IP network and demonstrate how a number of robust optimization schemes allow to find fixed MPLS configurations that are close to the performance limits given by time-varying routing under full demand knowledge. We present a novel scheme for computing a sparse MPLS mesh to complement a baseline routing, and explore how the performance depends on the size of the partial mesh. Corresponding methods for robust OSPF optimization are discussed and a number of challenges are detailed.

Ambient Networks Management Challenges and Approaches

System management addresses the provision of functions required for controlling, planning, allocating, monitoring, and deploying the resources of a network and of its services in order to optimize its efficiency and productivity and to safeguard its operation. It is also an enabler for the creation and sustenance of new business models and value chains, reflecting the different roles the service providers and users of a network can assume. Ambient Network represents a new networking approach and it aims to enable the cooperation of heterogeneous networks, on demand and transparently, to the potential users, without the need for pre-configuration or offline negotiation between network operators. To achieve these goals, ambient network management systems have to become dynamic, adaptive, autonomic and responsive to the network and its ambience. This paper discusses relationships between the concepts of autonomous and self-manageability and those of ambient networking, and the challenges and benefits that arise from their employment.

Towards ambient networks management

Ambient Networks (AN) are under development and they are based on novel networking concepts and systems that will enable a wide range of user and business communication scenarios beyond todays fixed, 3rd generation mobile and IP standards. Central to this project is the concept of Ambient Control Space (ACS) and the Domain Manager control function, which manages the underlying data transfer capabilities and presents a set of interfaces towards the supported services and applications. Network Management Systems of Ambient Networks must work in an environment where heterogeneous networks compose and cooperate, on demand and transparently, without the need for manual (pre or re)-configuration or offline negotiations between network operators. To achieve these goals, ambient network management systems must become dynamic, distributed, self-managing and responsive to the network and its ambience. This paper describes the different management research challenges and four complementary solution approaches (i.e. Pattern-based Management, Peer-to-Peer Management, (Un)PnP Management, Traffic Engineering Management Application Approaches) that enable efficient management of ambient networks, and the relationships between them, and presents the main results achieved so far.

Performance of traffic engineering in operational IP networks – an experimental study

Today, the main alternative for intra-domain traffic engineering in IP networks is to use different methods for setting the weights (and so decide upon the shortest-paths) in the routing protocols OSPF and IS-IS. In this paper we study how traffic engineering perform in real networks. We analyse different weight-setting methods and compare performance with the optimal solution given by a multi-commodity flow optimization problem. Further, we investigate their robustness in terms of how well they manage to cope with estimated traffic matrix data. For the evaluation we have access to network topology and traffic data from an operational IP network.

Robust load balancing under traffic uncertainty--tractable models and efficient algorithms

Routing configurations that have been optimized for a nominal traffic scenario often display significant performance degradation when they are subjected to real network traffic. These degradations are due to the inherent sensitivity of classical optimization techniques to changes in model parameters combined with the significant traffic variations caused by demand fluctuations, component failures and network reconfigurations. In this paper, we review important sources for traffic variations in data networks and describe tractable models for capturing the associated traffic uncertainty. We demonstrate how robust routing settings with guaranteed performance for all foreseen traffic variations can be effectively computed via memory efficient iterative techniques and polynomial-time algorithms. The techniques are illustrated on real data from operational IP networks.

Robust Routing Under BGP Reroutes

Configuration of the routing is critical for the quality and reliability of the communication in a large IP backbone. Large traffic shifts can occur due to changes in the inter-domain routing that are hard to control by the network operator. This paper describes a framework for modeling potential traffic shifts due to BGP reroutes, calculating worst-case traffic scenarios, and finding a single routing configuration that is robust against all possible traffic shifts due to BGP reroutes. The benefit of our approach is illustrated using BGP routing updates and network topology from an operational IP network. Experiments demonstrate that the robust routing is able to obtain a consistently strong performance under large inter-domain routing changes.

Access and Path Selection in Ambient Networks

As the choice of wireless network technologies increases and different wireless network technologies are connected together the bottleneck of a path will not necessarily be the first access link. Instead a link further upstream could be the bottleneck. This paper introduces a novel generalisation of access selection which takes properties beyond the first hop into consideration for the selection of path for communication. The motivation for design choices and the algorithms involved in the access and path selection are explained. In addition, a use case is included to describe the solution and illustrate its benefits.

Robust Load-balancing under Statistical Uncertainty: Models and Polynomial-time Algorithms

We study the problem of guaranteed-performance routing under statistical traffic uncertainty. Relevant traffic models are presented and a polynomial-time algorithm for solving the associated robust routing problem is given. We demonstrate how our techniques, in combination with fundamental limitations on the accuracy of estimated traffic matrices, enable us to compute bounds on the achievable performance of OSPF-routing optimized using only topology information and link count data. We discuss extensions to other types of traffic uncertainties and describe an alternative, more memory efficient, algorithm based on combined constraint and column generation. The proposed techniques are evaluated in several numerical examples to highlight the features of our approach.

Cautious weight tuning for link-state routing

Link-state routing protocols are widely used for intradomain routing in the Internet. These protocols are simple to administer and automatically update paths between sources and destinations when the topology changes. However, finding link weights that optimize network performance for a given traffic scenario is computationally hard. The situation is even more complex when the traffic is uncertain or time-varying. We present an efficient heuristic for finding link settings that give uniformly good performance also under large changes in the traffic. The heuristic combines efficient search techniques with a novel objective function. The objective function combines the network performance with a cost of deviating from desirable features of robust link weight settings. We assess performance of our method using traffic data from an operational IP backbone.