Forum: Reliability & Maintainability Questions and Answers
Topic: Reliability & Maintainability Questions and AnswersTopic Posted by: Reliability & Maintainability Forum (firstname.lastname@example.org )
Organization: System Reliability Center
Date Posted: Mon Aug 31 12:47:36 US/Eastern 1998
Original Message:Posted by: Alex Olson (email@example.com )
Date posted: Tue Jul 19 19:03:35 US/Eastern 2005
Subject: Markov Models for Relability
I'm trying to develop a model to measure the average execution time/efficiency of a large computer system under various fault-tolerance techniques (e.g. checkpointing). I came across one paper that used a Markov Cost model to model a repairable computer system. They used three states (start,failed segment,successful segment). I understand how they determined the probabilities and costs for each transition, but not the average cost of all paths. They derived the average execution time/efficiency from the average cost of the START->SUCCESS path. Of two other papers (that give the same result), this derivation seems cleaner [I wish to add more states]. Can anyone direct me to an online reference for this type of material?