Scalable Session Messages in SRM

Multi-party applications are of great importance, and perhaps the greatest challenge is achieving both a) robustness in terms of adaptation to dynamic topology and group membership, and b) scalability in terms of bandwidth, state, and processing, as the size of a group grows. Scalable Reliable Multicast (SRM) [1] is a rich example of a robust design intended to work across a wide range of group sizes and dynamic topologies. However, the adaptation mechanisms in SRM rely on shared group state achieved via exchange of session messages. Similar synchronization is likely to be of importance in other multiparty applications and services. Various mechanisms have been proposed to reduce the overhead of this loose group synchronization. This paper applies the concept of self-configuring hierarchy to SRM. Unlike previous proposals, our mechanism uses a stochastic algorithm for self-configuration based on randomized timers and local appropriateness measures. We present initial evaluations of the impact of this mechanism on SRM performance, and evaluate the hierarchical structure formed by the protocol participants. Many interesting questions remain to be investigate in future work such as self-evaluation of appropriateness, dynamics and stability of self-configuration process.