The invention provides an apparatus and method to establish media sessions for media streams crossing a network boundary. The system includes a media relay controlled by a media configurator control module. The media relay reserves media paths (that include ports in the network boundary), opens the media paths, closes the media paths, and provides information about the media paths. A media configurator is adapted to communicate with the media configurator control module and the media relay. The control module has an event handler handling multimedia session events, a local address resolver that determines if an address identifier of the media session belongs to a private address space and a control element used to establish the media path. The control element manages resources for the media relay. A state-refresh timer is used to maintain state consistency between all media relays controlled by a proxy engine and control elements.