[Hibernate] – How to work with locks ?

 

Problem :

Imagine the situation where in a store, you need to publish daily promotions and perform editing products. This store has several employees. What will happen if two or more of them update the information on a product at the same time? What data will be worth? What is saved first or last? As our application should handle this?

Answer: 

The data that will prevail will be the last informed. I call this : write conflict .

Solutions:

We can handle this scenario in some ways: A pessimistic approach would be to prevent the conflict from happening and ask for the database lock the record being updated.

Example:

Product product = em.find(Product.class, 1);
em1.lock(product,LockModeType.PESSIMISTIC_WRITE);

Probably, it will not be the best solution. If another user attempts to perform this operation will have no access to information because it will be blocked.

Lock Optimistic

The idea is that the application / database never lock the access to the record, being optimistic, assuming that this is a situation not so frequent. But what do we do when there is a concurrent update? We have to prevent and treat the way that is most convenient according to the application business.

By @Version

@Entity
public class Product {
    // ...
    @Version
    private Integer version; 
    // ...
}

To save the version, you can use a numeric field or a timestamp (Calendar or Date). If the entity has @Version throughout update done, hibernate will automatically check the value of this field. If the record in the database has a lower value than is being sent to the version field, it accepts the update and increases its value. If you have a higher value, an exception of StaleObjectStateException type will be triggered within a javax.persistence.OptimisticLockException.

So, if two people try to update the same record, the first person will upgrade increasing its value. The last will be prevented from updating as attempts to sent a lower value than in the database.