Evaluation of a server-based traffic engineering architecture suitable for large-scale MPLS networks

Internet traffic has continued to grow exponentially in recent years so that operators frequently face increasing congestion and complex problems in the operation of backbone networks. Efficient traffic engineering is one way of solving these problems. Generally, traffic-engineering functions are implemented per router. This is far from ideal in a large-scale network because of delays in synchronization and path set up. To solve these problems, we have proposed a scalable server-based traffic-engineering architecture that is suitable for large-scale networks. In this paper, we evaluate the scalability of traffic engineering architecture in terms of both route convergence time and path set up time. The evaluation shows that there is a delay in the convergence time of distributed architecture and that the path set up time for centralized architecture is faster than that of distributed architecture. Thus a centralized traffic engineering architecture could be applied to large-scale networks and this would lead to more efficient use of network resources.