Atomic Multicast harder than Atomic Broadcast

Atomic broadcast and atomic multicast are communication primitives which ensure that processes agree on the sequence of messages they deliver. Whereas atomic broadcast is used to send messages to the set of all the processes in the system, atomic multicast sends messages to speciic subsets of processes. We say that an atomic multicast algorithm is \genuine", if only the sender and the destination processes of a message take part in the protocol needed to deliver the message. We formally deene the genuine atomic multicast problem, and we show that this problem turns out to be harder than the atomic broadcast problem. Whereas atomic broadcast can be solved with unreliable failure detectors that can make false failure suspicions, we show that atomic multicast cannot.

[1]  Sam Toueg,et al.  Fault-tolerant broadcasts and related problems , 1993 .

[2]  Rachid Guerraoui,et al.  Total order multicast to multiple groups , 1997, Proceedings of 17th International Conference on Distributed Computing Systems.

[3]  Leslie Lamport,et al.  Time, clocks, and the ordering of events in a distributed system , 1978, CACM.

[4]  Sam Toueg,et al.  Unreliable failure detectors for reliable distributed systems , 1996, JACM.

[5]  Xiaohua Jia A Total Ordering Multicast Protocol Using Propagation Trees , 1995, IEEE Trans. Parallel Distributed Syst..

[6]  Hector Garcia-Molina,et al.  Ordered and reliable multicast communication , 1991, TOCS.