Data warehouses and data marts have long been considered as the unique solution for providing end-users with decisional information. More recently, data lakes have been proposed in order to govern data swamps. However, no formal definition has been proposed in the literature. Existing works are not complete and miss important parts of the topic. In particular , they do not focus on the influence of the data gravity , the infrastructure role of those solutions and of course are proposing divergent definitions and positioning regarding the usage and the interaction with existing decision support system. In this paper, we propose a novel definition of data lakes, together with a comparison with other over several criteria as the way to populate them, how to use, what is the Data Lake end user profile. We claim that data lakes are complementary components in decisional information systems and we discuss their position and interactions regarding the other components by proposing an interaction model.