In this tutorial we provide a comprehensive understanding of the problem of ontology-based data access, from both the theoretical and the practical points of view. We address several problems that are crucial in this context, such as expressiveness/efficiency tradeoff, query processing, impedance mismatch between ontology and data levels, and integration of multiple data sources. We present solutions to these problems based on recent research results in the area of tractable Description Logics, and we provide also a ``hands-on'' experience with QuOnto, a state-of-the-art system for ontology-based data access.