Markovian models for performance and dependability evaluation

  • Authors:
  • Boudewijn R. Haverkort

  • Affiliations:
  • RWTH Aachen, Aachen, Germany

  • Venue:
  • Lectures on formal methods and performance analysis
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Markovian models have been used for about a century now for the evaluation of the performance and dependability of computer and communication systems. In this paper, we give a concise overview of the most widely used classes of Markovian models, their solution and application. After a brief introduction to performance and dependability evaluation in general, we introduce discrete-time. Markov chains, continuous-time Markov chains and semi-Markov chains. Stepwisely, we develop the main equations that govern the steady-state and the transient behaviour of such Markov chains. We thereby emphasise on intuitively appealing explanations rather than on mathematical rigor. The relation between the various Markov chain types is explained in detail. Then, we discuss means to numerically solve the systems of linear equations (both direct and iterative ones) and the systems of differential equations that arise when solving for the steady-state and transient behaviour of Markovian models.