Distributed computing systems are attractive due to the potential improvement in availability, fault-tolerance, performance, and resource sharing. Modeling and evaluation of such computing systems is an important step in the design process of distributed systems. We present a two-level hierarchical model to analyze the availability of distributed systems. At the higher level (user level), the availability of the tasks (processes) is analyzed using a graph-based approach. At the lower level (component level), detailed Markov models are developed to analyze the component availabilities. These models take into account the hardware/software failures, congestion and collisions in communication links, allocation of resources, and the redundancy level. A systematic approach is developed to apply the two-level hierarchical model to evaluate the availability of the processes and the services provided by a distributed computing environment. This approach is then applied to analyze some of the distributed processes of a real distributed system, Unified Workstation Environment (UWE), that is currently being implemented at AT&T Bell Laboratories.>

A two-level hierarchical model is proposed to analyze the availability of distributed systems as perceived by their users. At the higher level (user level), the availability of the tasks (processes) is analyzed in terms of the availability of the system components. At the lower level (component level), detailed Markov models are developed to analyze the component availabilities. These models take into account the hardware/software failures, congestion and collisions in communication links, allocation of resources, and the redundancy level. Also presented is the availability analysis of some of the services provided by the unified workstation environment (UWE) currently being implemented at AT&T Bell Laboratories.>

The delay characteristics of the Datakit virtual circuit switch (VCS) are investigated, and its contribution to the overall end-to-end character echo response time delay on large mainframe UNIX (LMU) systems is evaluated. The end-to-end delay consists of the Datakit network, the device attached control unit (DACU), and host component delays. The echo response time is the amount of time it takes for a character to be echoed back by the host. To measure the end-to-end terminal-to-host delay and the Datakit contribution to the overall delay, two parallel experiments were set up. The first experiment provides direct dialing from a 3B2 computer into a LMU by passing Datakit, whereas the second test utilized the Datakit terminal-to-host network to access the LMU during the same time intervals. The experiments provide a considerable amount of data to evaluate the delay with and without the Datakit component. the 50th percentile echo response time clearly indicates that the Datakit contribution to the overall response time is around 50 ms with very little variance. The overall Datakit delay fluctuates between 35 ms and 70 ms.>

The B-ISDN (broadband integrated services digital network) is a virtual circuit-switched network that can be used for services requiring high-speed packet switching. Work is already underway to define and standardize B-ISDN in the CCITT (International Telegraph and Telephone Consultative Committee). The B-ISDN will use a flexible multiplexing format called ATM (asynchronous transfer mode). In the US proposal the B-ISDN is based on the SONET (synchronous optical network) specifications as a physical layer. An introduction is presented to the subject of computer network architecture in general, B-ISDN protocols, and interface structures, and some of the interfaces that can be used in establishing B-ISDN architecture are discussed. The authors also discuss the relationship between interfaces at reference points, the ATM header format, and SONET architecture.>

The authors propose models for defining the availability metric of a distributed computing environment. The metric is a meaningful indicator of a distributed computing environment's performance from the reliability point of view. The proposed models can also be used to efficiently design, engineer, and monitor a distributed computing environment's activities. The models are general and can be applied to any distributed computing environment for availability evaluation.>