Detecção dinâmica de antipadrões em sistemas baseados em serviços utilizando um sistema multiagente

Abstract

During development of Service-Based Systems (SBS), bad solutions, design or implementation errors can lead to antipatterns that, as opposed to patterns, are bad specifications for recurring problems. Antipatterns result in deviations of functionality during the execution of the system, not adequately satisfying its requirements, which can degrade the Quality of Service (QoS) of the SBS. Its detection and correction are of extreme importance, the emergence of antipatterns may hinder future maintenance and evolution of SBS. As detection method, this work proposes the use of a Multiagent System (MAS), executing search at runtime and notifying the system administrator of QoS drops on the SBS caused by antipatterns. The detection of these antipatterns allows the agents to act in correction the project using the solution recommended by the MAS for the detected problem. The results show that, in the worst cases, the emergence of antipatterns reached extrapolate up to twice the response time of services in a Web Service. The MAS was able to detect, in execution time, the 3 antipatterns studied at work. When adopted, the solutions proposed by the MAS also normalized the response time