Third International Conference on Performance of Distributed Systems and Integrated Communication Automated Generation and Analysis of Markov Reward Models Using Stochastic Reward Nets. 5.3.2 Non-homogeneous Ctmcs 5.2.1 Stiness Avoidance 5.2.2 Stiness Tolerance

In this tutorial, we discuss several practical issues regarding speci cation and solution of dependability and performability models. We compare model types with and without rewards. Continuous-time Markov chains (CTMCs) are compared with (continuous-time) Markov reward models (MRMs) and generalized stochastic Petri nets (GSPNs) are compared with stochastic reward nets (SRNs). It is shown that reward-based models could lead to more concise model speci cation and solution of a variety of new measures. With respect to the solution of dependability and performability models, we identify three practical issues: largeness, sti ness, and non-exponentiality, and we discuss a variety of approaches to deal with them, including some of the latest research e orts. This research was partially supported by the National Aeronautics and Space Administration under NASA Contract No. NAS1-19480 while the rst two authors were in residence at the Institute for Computer Applications in Science and Engineering (ICASE), NASA Langley Research Center, Hampton, VA 23681.