Relations in 1NF can have significant design flaws resulting in reduced efficiency. The goal of second normal form (2NF)
is to correct these flaws.
Consider the following table, which might be used to list cars available for sale at different dealerships. The relation assumes that a given car can only be stocked by a single dealer.
Note that, in this example, values in the CarID field are equivalent to book ISBN's.
The CarID denotes a particular make and model of car, but not a specific instance of that make and model.
In relational notation, this table would be written as:
Car (CarID, Color, Make, Model, Year, DealerID, DealerName)
The problem with this relation is that the fields relating to a car's dealer are not functionally dependent on the primary key (CarID). If any car can be acquired from more than one dealer, you will waste space by creating a second relation to save information about that car and its second dealer.
For a relation to be in 2NF, every descriptor (non-key column) must be functionally dependent
on every primary-key column in the relation.
The next lesson defines functional dependencies
The solution to anomalies in a first normal form relation is to break down the relation so there is one relation for each entity in the 1NF relation.
The orders relation, for example, will break down into four relations (customers, items, orders, and line items). Such relations are in at least second normal form (2NF).
In theoretical terms, second formal form relations are defined as follows:
- The relation is in first normal form.
- All non-key attributes are functionally dependent on the entire primary key.
The new term in the preceding is functionally dependent, a special relationship between attributes.
: A one-way relationship between two attributes so that, for any unique value of one attribute, there is only one value of the other attribute.