We present a delay analysis of three generic classes of multicast transport protocols. The first class considered is an unreliable scheme that works without retransmission of messages. This class also includes forward error correction approaches. The second class uses a positive acknowledgment and retransmission scheme to guarantee reliability. Finally, the third class is a hierarchical approach to avoid the well-known ACK implosion problem for large receiver groups.

Our results show that only the unreliable and the hierarchical protocol class provide scalability for large receiver groups. For delay sensitive applications we can conclude from the results that in case of low packet loss probabilities, reliable multicast protocols provide low average delays, which are only slightly increased compared to unreliable protocols. However, if we take the maximum delay, the delay is significantly increased.