Reliability Prediction and Sensitivity Analysis Based on Software Architecture

  • Authors:

  • Swapna S. Gokhale;Kishor S. Trivedi

  • Affiliations:

  • -;-

  • Venue:
  • ISSRE '02 Proceedings of the 13th International Symposium on Software Reliability Engineering
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Prevalent approaches to characterize the behavior ofmonolithic applications are inappropriate to model modernsoftware systems which are heterogeneous, and are builtusing a combination of components picked off the shelf,those developed in-house and those developed contractually.Development of techniques to characterize the behaviorof such component-based software systems based ontheir architecture is then absolutely essential. Earlier effortsin the area of architecture-based analysis have focusedon the development of composite models which are quitecumbersome due to their inherent largeness and stiffness.In this paper we develop an accurate hierarchical model topredict the performance and reliability of component-basedsoftware systems based on their architecture. This modelaccounts for the variance of the number of visits to eachmodule, and thus provides predictions closer to those providedby a composite model. The approach developed inthis paper enables the identification of performance and reliabilitybottlenecks. We also develop expressions to analyzethe sensitivity of the performance and reliability predictionsto the changes in the parameters of individual modules. Inaddition, we demonstrate how the hierarchical model couldbe used to assess the impact of changes in the workload onthe performance and reliability of the application. We illustratethe performance and reliability prediction as well assensitivity analysis techniques with examples.