We present a method to generate optical transport network topologies that preserve the major characteristics of
real-world transport network topologies. We also present an heuristic based on a genetic algorithm that,
combined with the topology generator, can be used to obtain the lowest cost topology in a multi-layer optical
transport network. In order to deal with survivable networks, we derive expressions to estimate the protection
and restoration coefficients, without complete information. Finally, the impact of the traffic randomness in the
obtained estimations is analyzed.