This paper describes the design of an object replication scheme for the Arjuna distributed system. ThedesignsupportsK-resiliency,where,intheabsenceofnetworkpartitions,Koutofa totalofK+1replicafailurescanbetoleratedbeforeanobjectbecomesunavailable.Thescheme chosen employs active replication where each and every functioning replica of an object carries out processing. Computations are structured as atomic actions (atomic transactions). The paper presents the details of how object groups are created and terminated, how a group can be invokedandobjectreplicasinsertedandremovedinaconsistentmannerinthepresenceofnode failures. Key words. Atomic actions, transactions, active replication, fault-tolerance, object oriented systems, distributed systems, replication. Appeared in the Proceedings of the First IEEE Workshop on Replicated Data, Houston, November 1990.
Mark C. Little, Santosh K. Shrivastava