Reuse JPA Entities as DTO

Note : Major design advantages of JPA Entities are they can detached and used across tiers and networks and later can by merged.

Checkout this new way of querying entities in JPA 2.0


String ql = "SELECT new prepclass2.Employee (e.firstname, e.lastname) FROM Employee e";
List<Employee> dtos = em.createQuery(ql).getResultList();

The above query loads all Employee entities but with subset of data i.e. firstname, lastname.

Employee entity looks like this.

@Entity
@Table(name="emp")
public class Employee implements Serializable {

    private static final long serialVersionUID = 1L;
    @Id
    @GeneratedValue(strategy = GenerationType.AUTO)
    private Long id;
    @Column
    private String firstname;
    @Column
    private String lastname;
    @Column
    private String username;
    @Column
    private String street;
    @Column
    private String city;
    @Column
    private String state;
    @Column
    private String zipcode;
    @Column
    private String country;

    public Employee(){}
    public Employee(String firstname, String lastname){..}
    .
    .
}


Benefits of the above approach

1. No separate DTO classes are required
2. Re-Use same Entities for exchanging data in any combinations
3. You can also create DTO classes and load data but why would you wanna do that when we can reuse entity classes
4. No need to write converters (i.e. DTO-Assembler) for copying data from entities to dtos and vise-versa
5. Performance efficient

Sample Load Test Results for loading 50K Employee entities using Hibernate JPA.


  General Way          -->  4621 ms
  Constructor Query -->   309 ms





Contructor Expression Rules

1. Entity or non-Entity class can be used to load coarse-grained data
2. Class should have a matching constructor defined with correct argument types

Also you can use Constructor queries with non-entity classes for instance

String ql = "SELECT new prepclass2.EmployeeDto (e.firstname, e.lastname) FROM Employee e";

I encourage you to write your own test cases to validate performance gains. though I'm more than convince this is a nice way of sharing required data between tiers in a more performance efficient and simple way without introducing more dto classes to digest.