Evolutionary algorithm for generating optimized configuration of computational distributed cluster considering simulation environment and specified workload

Today cloud computing provides effective solution for any business infrastructure problem through quick adaptation to any conditions and requests. However, testing the intended cluster configuration in a sandbox or using cloud resources is sometimes impossible in principle. In this situation arises the problem of choosing the right configuration of the computational system, taking into account the financial and operational criteria. This research proposes using a genetic algorithm scheme to generate a computational cluster configuration and to evaluate the quality of the created solutions through a simulated computing environment with integrated models and adapted case studies. This approach provides a fast selection of configurations even for complex configurations and surpasses heuristics that implement the basic logic of experts as well.