We explore the domain of realistic simulation of dynamically
changing networks. Throughout this exploration we define and implement a complete packet-level simulation framework and quickly manage
changes in a dynamic network. This is accomplished by our utilization of
the discrete event simulator OMNeT++ and our own non-trivial extensions of the simulation tool. Furthermore we also analyze and transform
real world traffic data from a large European ISP, leveraging the realism of the traffic data to improve the otherwise static traffic demands
of the Internet Topology Zoo. Finally we contribute with our own genetic algorithm Spungeet, as a means of swiftly reacting to the constant
changes of a parallel running network. Spungeet leverages the fact that
the order in which demands are routed makes a large difference in the
ability to deliver the packets and thus it is evident that mapping the
demands to individuals will enable quick generation of well performing
data planes for the network. The Spungeet algorithm tries to maximize
the percentage of packet delivery and performs well in both scenarios
with and without failures.