The relational data model and relational algebra

The early days of database engines (1960’s) saw several competing “data models” (the formal specifications for how the system represents and reasons about data). Academic researchers had proposed several models that were rich, expressive, and provided admirable data independence, but that were impossible to implement efficiently. The models used by working systems, on the other hand, tended to be very limited. The “hierarchical” and “network” models were two of the most popular. Both offered excellent runtime performance—when used properly—but left nearly everything up to the application developers unlucky enough to work with them. In particular, these simple models provided very little data independence, so any change to the way data was stored required corresponding changes to the applications using that data.