> Has anyone modeled what the multicast to multiple-unicast efficiency > threshold is? An interesting experiment to perform, without doubt. (Experiment would be more interesting than modelling.) -- Juliusz