Automatic performance space exploration of web applications using genetic algorithms

We describe a tool-supported performance exploration approach in which we use genetic algorithms to find a potential user behavioural pattern that maximizes the resource utilization of the system under test. This work is built upon our previous work in which we generate load from workload models that describe the expected behaviour of the users. In this paper, we evolve a given probabilistic workload model (specified as a Markov Chain Model) by optimizing the probability distribution of the edges in the model and generating different solutions. During the evolution, the solutions are ranked according to their fitness values. The solutions with the highest fitness are chosen as parent solutions for generating offsprings. At the end of an experiment, we select the best solution among all the generations. We validate our approach by generating load from both the original and the best solution model, and by comparing the resource utilization they create on the system under test.