Abordagem de análise do tempo de resposta para teste de desempenho em aplicações web

• Autor(a) principal: Guarienti, Priscila
• Data de Publicação: 2014
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da PUC_RS
• Texto Completo: http://tede2.pucrs.br/tede2/handle/tede/5278
• topic: INFORMÁTICA; ENGENHARIA DE SOFTWARE; SOFTWARE - AVALIAÇÃO; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO

Technology advances have allowed software systems to become more common in our society solving different types of problems we face everyday. This widespread use of software systems brought about some bigger and more complex solutions, which resulted in more systems that can fail. Furthermore, two important aspects in the software development process are being considered in every project: time to delivery and product quality. Since products have to be delivered faster, some faults may reside in the product and that could cause unforeseeable costs. In this scenario it is fundamental that the testing activity become one of the most important parts of the software development process to guarantee the quality of the final product. However, the testing activity is usually very expensive and time consuming, since most of the time it is executed manually. To improve that, Model Based Testing (MBT) has been used lately to generate testing artifacts automatically. These artifacts are usually extracted from the system model. This has helped to reduce the cost of testing, since the time to generate and execute tests can be reduced. Therefore, the main contribution of this work is to investigate how MBT can be applied for performance testing, i.e. how to include timing constraints in UML (Unified Modeling Language) diagrams and based on those constraints to execute the tests and verify whether the constraints are met or not. Another contribution of this work, is the setting of Service Level Agreements, i.e. set parameters to verify whether a system delivery answers based on pre-determined values. We applied our strategy to two sample applications, i.e. TPC-W and Skills, and measured their execution time using the LoadRunner Tool. Furthermore, we applied our whole strategy also to a complete case study, i.e. ChangePoint, and discuss some of our results.