As part of the European Strategic Programme for Research in Information Technology (ESPRIT), the Delta-4 project is seeking to define an open, faulttolerant, distributed computing architecture. This paper presents the overall Delta-4 framework for open, fault-tolerant, distributed computing systems and sketches the current implementation which is based on a local area network with specific atomic multicasting and error-processing protocols for communicating between replicated software components. Keywords: Fault-tolerant distributed systems, open distributed processing, replicated processing. atomic mulllast / I l l -I -: Abstract view of Delta-4 architecture 1.1.2 Fault assumptions. Two extreme assumptions about faulty hosts are allowed: fail-silent hosts and fail-uncontrolled hosts. The set of fault-tolerance techniques provided must allow either assumption to be made depending on the available host hardware and/or the criticality of the application. Intermediate host-Fail-silent e / , ' ' 'Dependable communication system m: Simplified Delta4 hardware architecture 2.1 Communication system The dependable communication system of the current Delta-4 implementation is called MCS (Multicast I;ommunicationSystem). MCS provides reliable multiendpoint communication based on multiendpoint connections [ IO]. Two sorts of multicasting are provided, termed visible and invisible multicasting. Visible multicasting pertains to the communication of data to one or more user-specified destinations. Invisible multicasting refers to implicif multiple destinations (or replicated destinations) resulting from the replication of software components for fault-tolerance. This section first overviews the layered structure of MCS and then details the atomic multicast protocol on top of which the higher-level services are built.2.1.1 MCS layered structure. Although work has commenced at within IS0 on multipoint communication [I 41, no standard broadcast protocols yet exist. Consequently, Delta4 proposes a hierarchy of specific protocols that are nevertheless in conformance with the ISO/OSI reference model. Despite the specificity of these protocols, there is a total compatibility with IS0 standards for the physical layer, with the services offered by the MAC sub-layer and by the application layer. In the present MCS implementation, both the presentation and network layers are null-layers.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.