Why you should never use the TABLE identifier generator with JPA and Hibernate


From a data access perspective, JPA supports two major types of identifiers:

  • assigned
  • generated

The assigned identifiers must be manually set on every given entity prior to being persisted. For this reason, assigned identifiers are suitable for natural keys.

For synthetic Primary Keys, we need to use a generated entity identifier, which is supported by JPA through the use of the @GeneratedValue annotation.

There are four types of generated identifier strategies which are defined by the GenerationType enumeration:

  • AUTO

The AUTO identifier generator strategy chooses one of the other three strategies (IDENTITY, SEQUENCE or TABLE) based on the underlying relational database capabilities.

While IDENTITY maps to an auto-incremented column (e.g. IDENTITY in SQL Server or AUTO_INCREMENT in MySQL) and SEQUENCE is used for delegating the identifier generation to a database sequence, the TABLE generator has no direct implementation in relational databases.

This post is going to analyze why the TABLE generator is a poor choice for every enterprise application that cares for performance and scalability.

Continue reading “Why you should never use the TABLE identifier generator with JPA and Hibernate”