On Service Models for Multicast Transmission in Heterogeneous Environments

We examine in this paper the tradeoff between application complexity, network complexity, and network effi- ciency. We argue that the design of the current Internet reflects a tradeoff between lower network complexity (no state in the network, no signalling) and higher application complexity (rate and error control mechanisms to obtain an adaptive application) assuming a unicast service model. For such a service model, a design methodology that leans heavily towards application com- plexity has proven very successful. However, we also argue that this tradeoff changes radically for a multicast/multilayer service model. There are several rea- sons for this. First, implementing a multicast/multilayer ser- vice requires per-flow state. The incremental cost of deploying a slightly more elaborate service model that takes into account flow dependence is much smaller than in the case of unicast. Second, several end-to-end functions, such as channel estima- tion and error control, are considerably more difficult for multi- cast/multilayer applications in a large-scale and heterogeneous environment than for unicast applications. Third, the operating point of a pure best-effort network is dictated largely by elastic applications (such as those based on TCP). Unfortunately, this operating point tends to be undesirable for multicast/multilayer applications, as they have for example to use FEC to protect high-priority layers. Other choices similarly lower the network efficiency. These insights motivate a new service model which slightly departs from the best-effort model, and which trades off a slightly higher network complexity for much lower application complexity and higher network efficiency. We describe this ser- vice model and the associated network protocols. The protocol complexity is only marginally higher than that of a simple multi- cast routing protocol with receiver-initiated join/leave capabili- ties. The dependencies between multilayer flows are established and maintained as soft state; therefore, no explicit session sig- nalling to establish and tear down flow dependence state is nec- essary.