Non-parametric Predictive Inference of Preventive Rejuvenation Schedule in Operational Software Systems

Abstract

In this paper we develop a novel approach to estimate the optimal preventive rejuvenation schedule which maximizes the steady-state system availability. In the case with unknown system failure time distribution, the preventive rejuvenation is triggered for the purpose of preventive maintenance of software system. We formulate the upper and lower bounds of the predictive system availability using the one-look ahead predictive survivor function, and derive the pessimistic and optimistic rejuvenation policies. In the real data analysis we focus on a real web server system and show the usefulness of the non-parametric predictive inference approach proposed in this paper.

An Adaptive Cost-Based Software Rejuvenation Scheme With Nonparametric Predictive Inference Approach

“…Even if T * up,n+1 differs from T − * low,n+1 , T * up,n+1 can be regarded as the optimal software rejuvenation schedule from a viewpoint of an absolute error average against the theoretical minimum cost. Rinsaka and Dohi [33] investigate the above NPI-based software rejuvenation policies under a different dependability criterion and apply the real data analysis of a web server system. x (1) x (2) x (3) x (4) x (5) x ( …”

“…More precisely, after the optimal preventive rejuvenation schedule T * n+1 (= T * up,n+1 ) for (n + 1)-st system failure is triggered, we predict the next software rejuvenation schedule for (n + 2)-nd system failure from n + 1 system failure data or a right-censored observation at T * up,n+1 . This problem has not been considered in Rinsaka and Dohi [33] and gives a useful prediction scheme of the software rejuvenation.…”

“…Then the key challenge is to develop a prediction-based adaptive software rejuvenation scheme. Rinsaka and Dohi [33,35] use a non-parametric predictive inference (NPI) approach by Coolen and Yan [10] and Coolen-Schrijner and Coolen [11]. They focus on a periodic software rejuvenation model Garg et al [20] and Suzuki et al [37] and derive the optimal NPI-based software rejuvenation time maximizing the steady-state availability criterion.…”

Dynamic Software Availability Model With Rejuvenation

“…Transaction-based software systems with queueing effects are considered by Garg et al [16], Okamura et al [31][32][33][34]. Rinsaka and Dohi [38,40] and Vaidyanathan and Trivedi [43] develop prediction-based approaches for triggering software rejuvenation. The above works deal with static optimization problems to determine the optimal software rejuvenation timing.…”

Non-parametric predictive inference of adaptive software rejuvenation schedule

“…In this paper we consider a basic two-step failure model with periodic rejuvenation considered by Suzuki et al [14] and Rinsaka and Dohi [11,12], and focus on the statistical estimation problem on the optimal periodic rejuvenation schedule maximizing the steady-state system availability. Main purpose of this paper is to develop a statistically non-parametric adaptive algorithm to estimate the optimal * The present research was partially supported by Grant-in-Aids for Scientific Research from the Ministry of Education, Sports, Science and Culture of Japan under Grant No.…”