Data Normalization Considered Harmful?
In relational databases, data normalization is a methodology that minimizes data redundancy to safeguard against logical and structural problems, such as update anomalies. Relational database normalization continues to be taught and practiced widely. Normalization was devised in the 1970s when the assumptions about data management and computer systems were very different from what they are today.
In this discussion we want to examine the problems associated with data normalization, such as complexity, performance, versioning, and the difficulty of mapping real-world business records to normalized data in a changing world. Also, the World Wide Web has impacted the creation and exchange of non-normalized business records.
When should and shoudn't data be normalized? When should alternative data representations, such as XML, JSON, and RDF be used to overcome normalization issues or to introduce schema flexibility?