A wonderful book from an academic point of view. It is mostly easy to read, yet it provides a solid knowledge about the theory that is behind relational databases. The book's approach is progressive: first it explains things in a more understandable way and then, when the reader has enough information, it proceeds to more formally correct explanations.
Unfortunately, the author's approach is a bit too theory-centric. He strongly criticises "inconsistencies" between the math and existing systems. This is a point of views that must be respected of course, but sometimes he doesn't see that these inconsistencies make them fast and able to scale as required by modern workloads.