Simple and stable dynamic traffic engineering for provider scale ethernet

Abstract

Trabalho apresentado no âmbito do Mestrado em Engenharia Informática, como requisito parcial para obtenção do grau de Mestre em Engenharia Informática The high speeds and decreasing costs of Ethernet solutions has motivated providers’ interest in using Ethernet as the link layer technology in their backbone and aggregation networks. Provider scale Ethernet offers further advantages, providing not only an easy to manage solution for multicast traffic, but also transparent interconnection between clients’ LANs. These Ethernet deployments face altogether different design issues, requiring support for a significantly higher number of hosts. This support relies on hierarquization, separating address and virtual network spaces of customers and providers. In addition, large scale Ethernet solutions need to grant forwarding optimality. This can be achieved using traffic engineering approaches. Traffic engineering defines the set of engineering methods and techniques used to optimize the flow of network traffic. Static traffic engineering approaches enjoy widespread use in provider networks, but their performance is greatly penalized by sudden load variations. On the other hand, dynamic traffic engineering is tailored to adapt to load changes. However, providers are skeptical to adopt dynamic approaches as these induce problems such as routing instability, and as a result, network performance decreases. This dissertation presents a Simple and Stable Dynamic Traffic Engineering framework (SSD-TE), which addresses these concerns in a provider scale Ethernet scenario. The validation results show that SSD-TE achieves better or equal performance to static traffic engineering approaches, whilst remaining both stable and responsive to load variations.

Original document

The different versions of the original document can be found in:

• http://hdl.handle.net/10362/4998