Advances in Distributed System Reliability
Suresh Rai, Dharma Prakash Agrawal · Medical Entomology and Zoology · 1990
Recent advances in VLSI circuitry have had tremendous impact on implementing a fairly complex process on a single chip. This development has led to increased use of stand-alone workstations connected in the form of a powerful distributed system. Potential benefits offered by such distributed systems include better cost performance resulting from exploiting parallelism in most of the algorithms, enhanced fault tolerance, a high degree of modularity, increased system throughput, and efficient sharing of resources. The authors focus on the reliability issues in such systems because the computation of system reliability metrics has become an integral part of the system designer's task. All of the three broad distributed system categories closely coupled, loosely coupled, and barely coupled types are considered. The text begins with an introduction to the topic and provides a critical assessment of current network models useful for evaluating reliability. The following features are addressed: multiprocessor and multiterminal reliability; multimode and dependent-failure analysis; performability analysis; task-based reliability and software reliability.