Copyright © 2008 The Institute of Electronics, Information and Communication Engineers
Regular Section -- Papers -- Dependable Computing |
Estimating Periodic Software Rejuvenation Schedules under Discrete-Time Operation Circumstance
1 The authors are with the Department of Information Engineering, Graduate School of Engineering, Hiroshima University, Higashihiroshima-shi, 739–8527 Japan. E-mail: dohi{at}rel.hiroshima-u.ac.jp, 2 The author is with the Department of Economic Informatics, Faculty of Economic Sciences, Hiroshima Shudo University, Hiroshima-shi, 731–3195 Japan.
 |
Abstract |
|---|
Software rejuvenation is a preventive and proactive solution that is particularly useful for counteracting the phenomenon of software aging. In this article, we consider periodic software rejuvenation models based on the expected cost per unit time in the steady state under discrete-time operation circumstance. By applying the discrete renewal reward processes, we describe the stochastic behavior of a telecommunication billing application with a degradation mode, and determine the optimal periodic software rejuvenation schedule minimizing the expected cost. Similar to the earlier work by the same authors, we develop a statistically non-parametric algorithm to estimate the optimal software rejuvenation schedule, by applying the discrete total time on test concept. Numerical examples are presented to estimate the optimal software rejuvenation schedules from the simulation data. We discuss the asymptotic behavior of estimators developed in this paper.
Key Words: rejuvenation, software aging, periodic preventive maintenance, expected cost, discrete-time models, non-parametric algorithms
Manuscript received March 24, 2006. Manuscript revised January 29, 2007.