| With the rapid development of computer technology, the applications of software in the national economy and social life are more and more widely and deeply. Its sales and usage are growing exponentially every year. Software applications have penetrated into our lives. However, with software growing in scale and the complexity enhancement dramatically, software failure may cause huge economic losses and even threaten human lives. Ensure the quality and reliability of software products has become the new issues in the development of software industry.Therefore, the software test has been paid more and more attention by software vendors as an important method to ensure the software quality and reliability. Because the software performance which is an important aspect of the software quality and reliability has been emphasized by software vendors and user, both software developers and end users all want to do a scientific performance evaluation to the system before it's on the line to reduce the performance risk.Software performance testing is an effective way to measure and test software performance. We can test the performance indicators of the system though simulating a variety of normal and abnormal peak load conditions. But the traditional performance test methods usually reflect the system performance only in a particular environment and time, the sustainability and extensibility can't be given a clear assessment and expectations. If the Software quality assessments haven't a clear guide, the tester will be lost when they face a series of test indexes. After the system updated, the previous performance testing data has no guidance to the follow-up testing. This way, often because of the project schedule, cost or other causes, the performance testing can't achieve the desired goals or be completed on time.This paper first introduced the theory and method of software test and mainly introduces the traditional software performance testing method and its features. Then it introduced the software reliability and reliability engineering research theory. Based on the analysis of the shortage of the traditional software performance test, put forward the improvement measures and improves the performance testing method in which the reliability engineering theory has been introduced. Though setting up the reliability model, introducing reliability engineering and defining the requirement of the system performance reasonably, the testing will be based on the reliability standards with a series of wise choices. Then we can establish the load model that corresponds to real environment and use scientific statistical analysis to get the performance test results. To do performance test in this method, the system can be effective self-assessment and take corrective measures to make the software reliability the largest optimized under the constraints of the project funding, schedule and resources.The major works are list below:1. Analyze the shortage of traditional software performance testing methods in software stability and reliability test. Put forward four big risks: performance index fuzzy; the test design setting less consideration; can't estimate test execution time; test data utilization rate is low and propose the improvement measures.2. Combine the characteristics of software reliability engineering and performance test, improve the traditional performance test methods and propose the performance test method for software reliability.3. Then used this method to test a communication recourse management system. Analyze system performance and reliability requirements, do the test, and make the test report, evaluate its performance reliability.4. Based on the failure data, using two kinds of main reliability growth model: Basic model and Logarithmic Poisson model to analyze and forecast the software reliability.5. Through analyze the test results, we can verify the validity of this method which is helpful to find out the high risk of performance bottlenecks and improve the system reliability. |
PDF Full Text Request