Predicting the software performance during feasibility study

Abstract

Performance is an important non-functional attribute to be considered for producing quality software. Software performance engineering (SPE) is a methodology having significant role in software engineering to assess the performance of software systems early in the lifecycle. Gathering performance data is an essential aspect of SPE approach. The authors have proposed a methodology to gather data during feasibility study by exploiting the use case point approach, gearing factor and COCOMO model. The proposed methodology is used to estimate the performance data required for performance assessment in the integrated performance prediction process (IP3) model. The gathered data is used as the input for solving the two models, (i) use case performance model and (ii) system model. The methodology is illustrated with a case study of airline reservation application. A regression analysis is carried out to validate the response time obtained in the use case performance model. The analysis shows the proposed estimation can be used along with performance walkthrough in data gathering. The performance metrics are obtained by solving the system model, and the behaviour of the hardware resources is observed. Bottleneck resources are identified and the performance parameters are optimised using sensitivity analysis.