What is Topology Management

This chapter discusses the use of peer-to-peer overlay networks to route around failures and congestion points in the Internet. The motivation for this type of overlay network is that often the network layer takes considerable time to react to link failure and is typically not capable of avoiding congestion. For instance, it has been reported that link failure may take the BGP system more than fifteen minutes to react. As a result, the services going over these links suffer from lost connectivity or severe quality degradation in the case of congestion. One technique to solve this problem is to use overlay routing techniques to detect these problems and to route around the problematic link(s). In this chapter we elaborate on the overall architecture of such a system. The concept is introduced and the architecture and implementation of prototype components implementing the behavior is discussed. Subsequently it is described how the overlay topology can be managed dynamically to achieve the same performance as a full mesh topology with a number of overlay edges that scales linearly with the number of nodes in the network.