Performability Analysis: Measures, an Algorithm, and a Case Study,

Abstract

Multiple-processor systems can provide higher performance and higher reliability/availability than single-processor systems. To properly assess the effectiveness of multi-processor systems, measures that combine performance and reliability are needed. The behavior of the multi-processor system is described as a continuous time Markov chain and associate a reward rate (performance measure) with each state. The distribution of performability for analytical models of a multi-processor system is evaluated using a recently improved polynomial-time algorithm that obtains the distribution of performability for non-repairable as well as repairable systems with heterogeneous components with a substantial speedup over earlier work. The system is analyzed with several Markov reward models is the (C.mmp) multi-processor system developed at Carnegie Mellon University. The example indicates that distributions of cumulative performance measures over finite intervals reveal behavior of multi-processor systems not indicated by either steady-state or mean values alone.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1988
Accession Number
ADA192477

Entities

People

  • A. V. Ramesh
  • K. S. Trivedi
  • R. M. Smith

Organizations

  • Yale University

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Air Force
  • Computational Complexity
  • Computational Science
  • Computer Science
  • Differential Equations
  • Eigenvalues
  • Equations
  • Fourier Series
  • Linear Systems
  • Partial Differential Equations
  • Probability
  • Random Variables
  • Reliability
  • Sequences
  • Steady State
  • Stochastic Processes
  • Time Intervals

Readers

  • Organizational Psychology.
  • Parallel and Distributed Computing.
  • Statistical inference.