When application developers use PHP and MySQL to build web interfaces or other data manipulation applications, they usually establish relations between tables, using the underlying SQL queries—for example, 'get an invoice and all its items' and 'get all books by an author'.

In the first versions of phpMyAdmin, MySQL was storing information about which table belonged to which database, but the relational data structure (how tables relate to each other) was not stored into MySQL. Relations were temporarily made by the applications to generate meaningful results. In other words, the relations were in our head.

This was considered a shortcoming of MySQL by phpMyAdmin developers and users, and so the team started to build an infrastructure to support relations. The infrastructure evolved to support a growing array of special features. We can describe this infrastructure as metadata (data about data).

phpMyAdmin 2.2.0 already had the bookmarks feature (being able to recall frequently...