Evaluation of multi‐tasking software system performance with consideration of module dependency

Abstract

This paper proposes new stochastic models that describe the task processing by the Markovian software availability model with perfect debugging and study the method to evaluate the system performance for the multi‐tasking software system having several modules. We assume that the software system, which is integrated by all modules, can process multiple tasks simultaneously. Each task, which is to be processed, is assumed to have multiple subtasks that are processed by the corresponding modules of the software system. Based on these models, we derive several performance measures, such as the expected numbers of subtasks and tasks whose processes can be completed within a certain time interval and the completion probability for each subtask and task. We also discuss the impacts that the number of modules constituting the system has on such performance measures. Furthermore, we consider the notion of module dependency for evaluation of certain probabilities of our interest and study its effects on such probabilities. Numerical examples are provided to illustrate our proposed models. Copyright © 2010 John Wiley & Sons, Ltd.

Performance Evaluation of Multi-Module Software System with Imperfect Debugging and Module Dependency

“…the number of modules in software system has 2 and 3. Lee et al [9] presents the result that the completion probability increases (decreases) as  increases from   to   if  is even(odd), regardless of the value of . However, throughout Table 1, we show the consistent result that the task completion probability with a positive  is always higher than the probability with a negative  in all 's.…”

“…Lee and Park [3] have proposed a Markovian imperfect debugging model for which the software failure is caused by two types of faults, one which is easily detected and the other which is difficult to detect. Tokuno and Yamada [4] consider a model for the multi-task processing system and provides a task Lee et al [9] suggest the concept of module dependency that the processing times of module tasks in a software task are mutually dependent and extends the task processing model of Tokuno and Yamada [4] to the model of a structure-based software with perfect debugging. The result shows that the task completion probability with positive(negative) module dependency is larger(smaller) than the task completion probability without module dependency when the number of modules within software is even.…”

“…But, the result is reversed when the number of modules within the software is odd. Therefore, it is difficult to use the result of Lee et al [9] when software developers newly upgrade the software by adding a new module with the module dependency that is the same as that of the existing software.…”

A Bayesian Approach Based on Bayes Minimum Risk Decision for Reliability Assessment of Web Service Composition

“…Web services are service-oriented architecture technologies that are executed using standard web protocols to ensure their operation varying platforms. With the continuous maturity of service-oriented technology and the promotion of service-oriented architecture (SOA) [1], multiple atomic web services have been implemented to achieve more powerful service compositions while satisfying relevant rules and addressing the demands of businesses. To maintain the quality of web service composition and utilize testing resources reasonably, assessment methods for web service compositions are required to exhibit high error detection capabilities, low cost consumption, and broad applicability [2].…”

“…When the number of test cases is very high or the test data are limited, software input is selected using statistical methods to promote rapid reliability growth of the software to ensure high reliability software demonstration testing [3]. In this context, Bayes reliability inference has been extensively applied to reliability estimation, and a wealth of experts and prior knowledge about system/subsystem performances are associated with it [1]. The Bayesian approach has been adopted because it combines subjective judgment or prior experience with data obtained from test samples.…”