Reliability Evaluation of Fault-Tolerant Multiprocessor Systems
Abstract
Under the auspices of this grant, we have developed a hierarchical, combinational-Markov approach for solving large reliability/availability/ performance models of systems. The approach allows the modeler to combine good aspects of both combinational models and Markov models to obtain a cost-effective solution to large models. The approach has been used in two Ph.D. dissertations (LTCL Jim Blake, who studied the performability of multiprocessor interconnection networks and Malathi Veeraraghavan, who modeled many fault-tolerant systems, including Boeing's IAPSA, Draper Laboratories' AIPS). Jim Blake has joined the Army AIRMICS Laboratory and Malathi has joined AT&T Bell Laboratories in Columbus. Other methods of dealing with complex system models that we have explored include automated methods of Markov model generation using Stochastic Petri nets. Efficient methods of solving stochastic Petri net models are being investigated by Gianfranco Ciardo in his dissertation. His work also involves applying SPN techniques for the performance analysis of concurrent programs. Much of our research deals with the transient solution of large and stiff Markov and Markov reward models.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 08, 1988
- Accession Number
- ADA205060
Entities
People
- Kishor Trivedi
Organizations
- Duke University