The Big Data era is an inevitable consequence of our capacity to generate and collect data. Therefore, data sources got a more dynamic behavior and the challenge to manage data source’s integration process and the network traffic between data producers and consumers has been overwhelmed by the number of data sources and their content’s volume, by the variety of their structures and formats and by the velocity of their appearance. This work presents a method to help users in choosing an appropriate approach (materialization or virtualization) for each data source in an integration environment.