Multicast forwarding plane in future networks: Source routing has a competitive edge

Abstract

Many applications in the current Internet follow the multicast paradigm, not the unicast one that has traditionally ruled the design of TCP/IP. These applications are, therefore, required to construct overlay networks for multicasting, which unfortunately offers too little efficiency to support the immanent emergence of trillions of groups. IP multicast, which is the natural implementation of the multicast paradigm, has not been deployed widely. Though recent years have seen several alternatives, we still have no clear direction for future networks that provide the multicast functionality efficiently. The aim of this paper is to indicate the most promising research direction on the multicast forwarding plane of future networks. Through intensive analysis, we conclude that packet-driven forwarding (source routing) has several advantages over traditional table-driven forwarding including IP multicast.