What is data normalization in a relational database?

Normalization is the process of structuring a relational database to reduce data redundancy and improve data integrity.

Main goals:

• Reducing redundancy: Avoid data duplication.
• Improving integrity: Ensure data consistency during insertions, deletions, and updates.

Normalization is based on the concept of normal forms (NF). The most common are:

• 1NF (First Normal Form):
  • Atomic values in each column.
  • Each column contains a single value of one type.
• 2NF (Second Normal Form):
  • It is in 1NF.
  • Every non-key attribute depends fully on the entire primary key. (For tables with a simple key, 2NF coincides with 1NF).
• 3NF (Third Normal Form):
  • It is in 2NF.
  • No transitive dependencies: a non-key attribute should not depend on another non-key attribute.

Example of normalization (transition from unnormalized form to 1NF):

Unnormalized table:

Order_ID	Customer	Products
1	Ivan	Milk, Bread
2	Maria	Apples, Bananas

Table in 1NF:

Order_ID	Customer	Product
1	Ivan	Milk
1	Ivan	Bread
2	Maria	Apples
2	Maria	Bananas

Higher normal forms (BCNF, 4NF, 5NF) exist and are used for more specific scenarios. The choice of normalization level depends on performance requirements and database complexity. Too high a normalization level can increase the number of joins and reduce read performance.