Topology Aggregation in PNNI Networks Part II: Node Aggregation

This paper and its part I (link aggregation) companion present our work done in the deployment of topology aggregation techniques in the layer 2 infrastructure of AT&T's Global Network to deal with network growth. In this paper, use of PNNI node aggregation for new node additions is examined. Two techniques are compared: simple and complex node representations. There is a tradeoff between accuracy in preserving the original topological features and compactness in representation to achieve scaling. While complex node is more accurate, it may require the use of link aggregation. Simple node scales well, but it may attract transit traffic. Our analysis of simple node suggests its use in offices with no major impact on latency performance. We illustrate with an application to regional hub offices with lots of links.