Enhancing discrete event network simulators with analytical network cloud models

Discrete event simulation of computer networks has signif-icant scalability issues, which makes simulating large scale networks problematic. We propose to extend the approach by combining it with analytical models for network clouds, which are much more efficient, if less accurate, than node-by-node models. Thus, simulation scenarios containing several ISP networks become feasible. We introduce a set of suitable analytical network cloud models and also de-scribe how these models can be implemented in the ns-2 simulator using a hot-plug mechanism.