The evolution of the vast amount of heterogeneous data sources is leading to the emergence of several new concepts. One of the best-known concepts that is emerging as a new and trending topic in the big data space is the data lake. This is a central repository that stores heterogeneous data sources in their native format, without any predefined schema. In the absence of an enforced schema, effective metadata management based on metadata models remains an active research topic to address the problems associated with the data lake: the "data swamp". The analysis of existing metadata models shows that there is no comprehensive model among them. In this paper, we present a generic and scalable metadata model, which refers to the ability to dynamically provision computing resources based on demand and to resize resources as needed during metadata integration. Our approach will be based on a functional architecture of the data lake, along with a set of features that promote the generality of the metadata model.