Integrating group communication with transactions for implementing persistent replicated objects

A widely used computational model for constructing fault-tolerant distribut ed applications employs atomic transactions for controlling operations on p ersistent objects. There has been considerable work on data replication tec hniques for increasing the availability of persistent data that is manipula ted under the control of transactions. Process groups with ordered group co mmunications (process groups far short) has also emerged as a model for bui lding available distributed applications. High service availability can be achieved by replicating the service state on multiple processes managed by a group communication infrastructure. These two models are often seen as ri vals. This paper examines whether a distributed transaction system can prof it from process groups for supporting replication of objects. A general mod el of distributed objects is used to investigate how objects can be replica ted for availability using a system that supports transactions (but no proc ess groups) and a system that supports process groups (but no transactions) . A comparative evaluation reveals how a distributed transaction system can exploit group communications for obtaining a flexible approach to supporti ng replication of objects.