Enterprise Java

JPA 2 | Dynamic Queries Vs Named Queries

JPA has its own Query language called JPQL. JPQL is very similar to SQL, with one major difference being that JPQL works with entities as defined in the application whereas SQL works with table and column names as defined in the database. JPA provides us with a variety of options when it comes to defining the JPA queries that will perform CRUD operations on our defined Entity classes.  These options are dynamic queries, Named Queries and Criteria queries. This post tries to look at each of the option in slight detail with the focus on when to use each type of query definition, what the performance problems might be and also some security threat associated with Dynamic Queries.

Dynamic Queries

Queries that are created at runtime by the application are called dynamic queries. Dynamic queries are created when we pass a simple JPA compliant query string to the createQuery method of the EntityManager class. Defining a dynamic query has its advantages and disadvantages. Lets look at each one of them in turn.

Advantages

The main advantage of using dynamic queries is in situation where you dont know how the query will look like until runtime and when the structure of the query is dependent on the user inputs or other conditions.

Disadvantages

The main disadvantage of dynamic queries is the cost associated with translating a JPQL query to SQL every time it is invoked. Most of the providers may try to cache SQL query generated from dynamic queries, but may not always be successful in doing that. Case where the query will not be cached or provider will find it difficult to cache is when concatination is used to bind parameter values directly into the query string. Hers an example where the query will not be cached because every time the JPQL is invoked a new query may be generated because of dynamic parameters.

@Stateless 
public class DynamicQueriesExample implements DynamicQuery {  
    @PersistenceContext(unitName="dq") 
    EntityManager em; 

    public long queryFinancialRecordsOfDept(String deptName, String companyName) { 
        String query = "SELECT d.records " + 
                       "FROM Department d " + 
                       "WHERE d.name = '" + deptName +  
                       "' AND " + 
                       "      d.company.name = '" + companyName + "'"; 
        return em.createQuery(query, Long.class).getSingleResult(); 
    } 

} 

In the above query, we are concatinating the deptName and companyName values in the query String. As a result, every time the method queryFinancialRecordsOfDept is invoked, a new query is generated. This new query is difficult to cache because the variables make the String unique almost every time. Thus if you have many dynamic queries in your application, and they are being called quite often, then you will have performance problems to solve.

The second problem with dynamic queries written like above is the actual concatenation operation. Since you are using simple String concatenation, and since Strings are immutable, the JVM will generate many String objects, most of which will be discarded in the end and will be lingering in your memory until the time the next garbage collection happens. This again may have affect on the performance of your application.

The third problem with dynamic queries, written like above, is that of security. In the above query for example, a hacker can easily pass in a value for companyName to alter the query to his advantage. Finding out what query in an application is expecting is easier than we think. Simple stack trace from the application reveals more than one can imagine.

So, if an application is expecting the company name to be specified at runtime by its user, a hacker in the above case can pass the value for companyName parameter as a GET or POST request with value of companyA OR d.company.name = companyB:

"SELECT d.records " + 
"FROM Department d " + 
"WHERE d.name = 'deptA' +  
   "AND d.company.name = 'companyA'
       OR d.company.name = 'companyB';

This type of security risk can easily be reduced by using named/positional Parameters feature of JPA. Named parameters help us in binding the values to the query at  a much later stage in the query processing. By using named parameters, the query is not altered every time for different parameters. Thus the query remains same and becomes easy for the provider to cache it.

The second advantage of using Named/Positional Parameters is that they are marshalled in to the query using the JDBC API and happens at the DB level and at DB level, the DB normally quotes the text passed as parameter. Thus, in the above case, we can alter the query to use named/positional parameters:

WAY 1 : Named Parameters

public long queryFinancialRecordsOfDept(String deptName, String companyName) { 
        String query = "SELECT d.records " + 
                        "FROM Department d " + 
                         "WHERE d.name = :deptName +  
                          "AND d.company.name = :compName;
        return em.createQuery(query,Long.class)
                 .setParameter("deptName" , deptName)
                 .setParameter("compName" , companyName)
                 .getSingleResult(); 
    } 

}

WAY 2 : Positional Parameters

public Long queryFinancialRecordsOfDept(String deptName, String companyName) { 
        String query = "SELECT d.records " + 
                        "FROM Department d " + 
                         "WHERE d.name = ?1 +  
                          "AND d.company.name = ?2; 
        return em.createQuery(query,Long.class)
                 .setParameter(1 , deptName)
                 .setParameter(2 , companyName)
                 .getSingleResult(); 
    } 

}

Way 1 of specifying the Parameters uses named variables that can be supplied values using the setParameter method on the Query Object.

Way 2 of specifying the Parameters uses numbers or indexes to bind query parameters to the query string.

Side Note: We can use the same named parameter multiple time in the query, but needs to bind the value only once using the setParameter function.

Named Queries

Named queries are a way of organising your static queries in a manner that is more readable, maintainable as well as performant.

Named queries in JPA are defined using the @NamedQuery annotation. This annotation can be applied only at class level and the Entity on which the Query will work is a good place to define Named Queries. For example, if a named query findAllItemRecords is defined to find all the Item entities in the Database table Item, then the Named query is normally defined  on the Item Entity. Here is an example:

@NamedQuery(name="Item.findAllItemRecords" , 
                      query="SELECT item " +
                                "FROM Item item")
@Entity
public class Item {

   @Id
   @Column(name="item_id")
   private String itemId;

   @Column(name="item_type")
   private String itemType;

   //.......
 }

One thing to notice above is that we are using concatenation operation on our string. But this will not be a performance problem like it is with dynamic queries because the Persistence Provider will convert the named queries from JPQL to SQL at deployment time and will cache them for later use. This means that the overhead of using concatenation is felt only at deployment time and not every time the query is used by the application. And the good thing about concatenating the query like above is that it makes it more readable and thus maintainable.

One thing to keep in mind with named queries is that the name of the query is scoped to be the entire Persistence Unit and thus it is not possible to have two named queries with the same name. You should use a qualifier for each named query. Normally using the name of the entity, like we have done in the above example, as prefix is a good choice.

We can define more than one NamedQueries for a given entity using @NamedQueries annotation. Lets have a look at an example of specifying multiple named queries.

@NamedQueries({
@NamedQuery(name="Item.findAllItemRecords" , 
                      query="SELECT item " +
                                "FROM Item item "
                                "WHERE item.itemId=:itemId),
@NamedQuery(name="Item.findItemByType" , 
                      query="SELECT item " +
                                "FROM Item item "
                                "WHERE item.itemType=:itemType)
})
@Entity
public class Item {

   @Id
   @Column(name="item_id")
   private String itemId;

   @Column(name="item_type")
   private String itemType;

   //.......
 }

We can use named queries in our method using the createNamedQuery method on the EntityManager.

public Item findAllItemRecords(String itemId) { 
         return em.createNamedQuery("Item.findAllItemRecords",  
                                    Item.class) 
                  .setParameter("itemId", itemId) 
                  .getSingleResult(); 

Summary

We discussed in this small blog post difference between Dynamic and Named queries in JPA. In the next blog post we will look at Criteria APIs and how they are used.

The contents in this blog post are a result of reading the excellent book Pro JPA 2 . I would recommend it to anyone who is working on JPA related projects.
 

Reference: JPA 2 | Dynamic Queries Vs Named Queries from our JCG partner Anuj Kumar at the JavaWorld Blog blog.

Anuj Kumar

Anuj is working as a Senior S/W Engineer in Netherlands. He is a Sun Certified Java Professional with experience in design and development of mid/large scale Java applications. He is the creator of EasyTest Framework(https://github.com/EaseTech) which is a Data Driven Testing Framework for Java.
Subscribe
Notify of
guest

This site uses Akismet to reduce spam. Learn how your comment data is processed.

0 Comments
Inline Feedbacks
View all comments
Back to top button