General framework for fault tolerance from ISO/ITU Reference Model for Open Distributed Processing (RM-ODP)

The ISO Reference Model for Open Distributed Processing (RM-ODP) is a software engineering enabler for the software architecture of distributed systems. It addresses the non-functional properties of distributed systems in terms of a framework of concepts, structuring rules, semantics and mechanisms. One such property is fault tolerance (FT). This paper provides a view of how RM-ODP addresses the FT framework. RM-ODP defines a model of FT and different recovery mechanisms that include recovery of an object, a group of objects, an interface, a binding, a bound interface, etc. Mechanisms for FT include a distinguished FT control object for the management of FT, replication, checkpoint/recovery, relocation, migration, and several transparencies to hide the FT mechanisms from the application developer. System-defined policies and schemata drive the FT. A brief comparison of this with the emerging proposal for a fault-tolerant version of the Common Object Request Broker Architecture (CORBA) is also provided.