The basic component of the JUMP model is a data unit.

A data unit is based on the Physical Data Model, and like a PDM each class box represents a physical data table.

Viewing a unit of data from a metadata perspective we look beyond the fields of data comprising each record and instead capture the column names as objects. Each row in a table represents a set of related data, and every row in the table has the same metadata structure as defined by the column names. The entire table represented in the data unit is comprised of the same metadata column names. The data unit views all data records in the table collectively as using the same metadata schema.

Data units are assembled into a metamodel. They are stacked unit-by-unit, or table-by-table, in a linearized readable order like blocks to form a data dictionary.