Advantages & Disadvantages of Normalizing a Database

Computer databases are everywhere, from those used by banks to track customer accounts to those used by websites to store content. Databases work best when they are designed well. Normalizing a database means to design the database structure to store data in a logical and related way. It is usual for all databases to be normalized, and normalizing a database has advantages and disadvantages.

Databases can hold a significant amount of information, perhaps millions or billions of pieces of data. Normalizing a database reduces its size and prevents data duplication. It ensures that each piece of data is stored only once.

Application developers who create applications to "talk" to a database find it easier to deal with a normalized database. The data they access is organized more logically in a normalized database, often similar to the way in which the real-world objects that the data represent are organized. That makes the developers' applications easier to design, write and change.

Referential integrity is the enforcement of relationships between data in joined tables. Without referential integrity, data in a table can lose its link to other tables where related data is held. This leads to orphaned and inconsistent data in tables. A normalized database, with joins between tables, can prevent this from happening.

A highly normalized database with many tables and joins between the tables is slower than a database without those attributes. Many people using a normalized database at the same time also can slow down database speed. In some cases, a certain amount of denormalization of the database may be required to improve database speed.

Normalizing a database is a complex and difficult task. Large databases with considerable amounts of information, such as ones run by banks, require careful analysis and design before they are normalized. Knowing the intended use of a database, such as whether it should it be optimized for reading data, writing data or both, also affects how it is normalized. A poorly normalized database may perform badly and store data inefficiently.