Solving Atomic Multicast When Groups Crash

In this paper, we study the atomic multicast problem, a fundamental abstraction for building fault-tolerant systems. In the atomic multicast problem, the system is divided into non-empty and disjointgroups of processes. Multicast messages may be addressed to any subset of groups, each message pos-sibly being multicast to a different subset. Several papers previously studied this problem either in localarea networks [3, 9, 20] or wide area networks [13, 21]. However, none of them considered atomicmulticast when groups may crash. We present two atomic multicast algorithms that tolerate the crash ofgroups. The first algorithm tolerates an arbitrary number of failures, is genuine (i.e., to deliver a messagem, only addressees of m are involved in the protocol), and uses the perfect failures detector P . We showthat among realistic failure detectors, i.e., those that do not predict the future, P is necessary to solve gen-uine atomic multicast if we do not bound the number of processes that may fail. Thus, P is the weakestrealistic failure detector for solving genuine atomic multicast when an arbitrary number of processes maycrash. Our second algorithm is non-genuine and less resilient to process failures than the first algorithmbut has several advantages: (i) it requires perfect failure detection within groups only, and not across thesystem, (ii) as we show in the paper it can be modified to rely on unreliable failure detection at the cost ofa weaker liveness guarantee, and (iii) it is fast, messages addressed to multiple groups may be deliveredwithin two inter-group message delays only.