Under the influence of Big Data, the information system must adapt and evolve. This development requires a transformation of decision-making systems to take into account the multiplication of data sources and the emergence of new analysis needs. While the issue of model evolution in Data Warehouses (DW) is not new, designing scalable DWs remains a challenge. In this article we propose an agile DW model based on the Anchor Modeling technique that allows its evolution. We define rules for moving from a multidimensional star model to an anchor model, then from an anchor model to a relational physical model. We validate our approach with a software prototype and compare our agile multidimensional model with traditional multidimensional star models.