Question

user(m:m)books

@Entity
public class User{
    @Id
    public int id;
    @ManyToMany(fetch = FetchType.LAZY, targetEntity = Book.class)
    @JoinTable(name = "user_book", joinColumns = @JoinColumn(name = "user_id"), inverseJoinColumns = @JoinColumn(name = "book_id"))
    @Fetch(FetchMode.SUBSELECT)
    public List<Book> books;
}

@Entity
public class Book{
    @Id
    public int id;
    ...
}


loadAllActiveUsers()
foreach user
   do(user.books)

假设我们有1000个用户，每个用户都拥有所有书籍（我们有1000本书）。当调用user.books时，hibernate会触发查询：

select book.* from  association inner join books b on b.id = association.book_id where association.user_id in (select u.id from users u)

此查询将包含1M行的结果，尽管我们有1000本书。这是因为hibernate由user_id而不是book_id加载。

如果hibernate将在where子句中进行关联的连接，那么结果将限制为不同的书籍，我们将不会查询重复的书籍。

select book.* from books where b.id in (select association.book_id from users u join association a on a.user_id = u.id)

有没有办法告诉hibernate以2个步骤进行子选择：获取关联的id而不是获取关联的实体？

由于冗余负载，hibernate subselect性能问题

0 个答案: