Case study on performance of a Group Communication System in the realm of voice over IP

To achieve availability and reliability, distributing data over several nodes within a network is widely used today. But distributing data to all nodes within a network becomes complicated if data should stay consistent while all of the nodes are allowed to insert, update or delete data. Common approaches focus on solutions where applications rely on replication algorithms and mechanisms provided by database systems. Instead the authors take another approach where the distribution of data is performed before the data is actually committed to the database system and thus, freeing the database system from the task of replication. To do so, a system that provides reliable data distribution and ordering mechanisms is necessary, placed between the application and the database system. In this paper we present a prototype of a redundant SIP-server that uses a Group Communication System to distribute data wherein the Group Communication System takes care of reliably delivering transactions to the nodes of a distributed database system in the same order. The implementation is explained and the corresponding results of a few test cases are presented.