Cloud computing is attractive to business owners and allows enterprises to start from the small and increase resources only when there is a rise in service demand, but cloud may become expensive. Fog computing has many advantages, and it is suited for the applications whereby real time is very important, but fog resources may also be highly limited. The cloud and fog computing may perform tasks together to attend different types of applications and mitigate their limitations. However, taking into account variables such as latency, workload and computational capacity, it becomes complex to define under what circumstances it is more advantageous to use the cloud layer or the fog. This paper proposes a stochastic Petri net to model such a scenario by considering cloud and fog. The model permits to configure 12 parameters including, for example, the number of available resources, workload and mean requests arrival time. We present a case study using a classical big data algorithm to validate the model. The case study is a practical guide to infrastructure administrators to adjust their architectures by finding the trade-off between cost and performance.