hibernate-orm/reference/en/modules/best_practices.xml

<chapter id="best-practices" revision="2">
    <title>Best Practices</title>

    <variablelist spacing="compact">
        <varlistentry>
            <term>Write fine-grained classes and map them using <literal>&lt;component&gt;</literal>.</term>
            <listitem>
                <para>
                    Use an <literal>Address</literal> class to encapsulate <literal>street</literal>,
                    <literal>suburb</literal>, <literal>state</literal>, <literal>postcode</literal>.
                    This encourages code reuse and simplifies refactoring.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Declare identifier properties on persistent classes.</term>
            <listitem>
                <para>
                    Hibernate makes identifier properties optional. There are all sorts of reasons why
                    you should use them. We recommend that identifiers be 'synthetic' (generated, with
                    no business meaning) and of a non-primitive type. For maximum flexibility, use
                    <literal>java.lang.Long</literal> or <literal>java.lang.String</literal>.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Place each class mapping in its own file.</term>
            <listitem>
                <para>
                     Don't use a single monolithic mapping document. Map <literal>com.eg.Foo</literal> in 
                     the file <literal>com/eg/Foo.hbm.xml</literal>. This makes particularly good sense in 
                     a team environment.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Load mappings as resources.</term>
            <listitem>
                <para>
                    Deploy the mappings along with the classes they map.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Consider externalising query strings.</term>
            <listitem>
                <para>
                    This is a good practice if your queries call non-ANSI-standard SQL functions. 
                    Externalising the query strings to mapping files will make the application more portable.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Use bind variables.</term>
            <listitem>
                <para>
                     As in JDBC, always replace non-constant values by "?". Never use string manipulation to 
                     bind a non-constant value in a query! Even better, consider using named parameters in
                     queries.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Don't manage your own JDBC connections.</term>
            <listitem>
                <para>
                    Hibernate lets the application manage JDBC connections. This approach should be considered 
                    a last-resort. If you can't use the built-in connections providers, consider providing your 
                    own implementation of <literal>net.sf.hibernate.connection.ConnectionProvider</literal>.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Consider using a custom type.</term>
            <listitem>
                <para>
                    Suppose you have a Java type, say from some library, that needs to be persisted but doesn't 
                    provide the accessors needed to map it as a component. You should consider implementing
                    <literal>net.sf.hibernate.UserType</literal>. This approach frees the application
                    code from implementing transformations to / from a Hibernate type.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Use hand-coded JDBC in bottlenecks.</term>
            <listitem>
                <para>
                    In performance-critical areas of the system, some kinds of operations (eg. mass update /
                    delete) might benefit from direct JDBC. But please, wait until you <emphasis>know</emphasis> 
                    something is a bottleneck. And don't assume that direct JDBC is necessarily faster. If need to 
                    use direct JDBC, it might be worth opening a Hibernate <literal>Session</literal> and using that SQL 
                    connection. That way you can still use the same transaction strategy and underlying connection 
                    provider.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Understand <literal>Session</literal> flushing.</term>
            <listitem>
                <para>
                    From time to time the Session synchronizes its persistent state with the database. Performance will
                    be affected if this process occurs too often. You may sometimes minimize unnecessary flushing by 
                    disabling automatic flushing or even by changing the order of queries and other operations within a 
                    particular transaction.      
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>In a three tiered architecture, consider using <literal>saveOrUpdate()</literal>.</term>
            <listitem>
                <para>
                    When using a servlet / session bean architecture, you could pass persistent objects loaded in
                    the session bean to and from the servlet / JSP layer. Use a new session to service each request. 
                    Use <literal>Session.update()</literal> or <literal>Session.saveOrUpdate()</literal> to update the 
                    persistent state of an object.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>In a two tiered architecture, consider using session disconnection.</term>
            <listitem>
                <para>
                    Database Transactions have to be as short as possible for best scalability. However, it is often
                    neccessary to implement long running Application Transactions, a single unit-of-work from the
                    point of view of a user. This Application Transaction might span several client requests and
                    response cycles. Either use Detached Objects or, in two tiered architectures, simply disconnect
                    the Hibernate Session from the JDBC connection and reconnect it for each subsequent request. Never
                    use a single Session for more than one Application Transaction usecase, otherwise, you will run
                    into stale data.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Don't treat exceptions as recoverable.</term>
            <listitem>
                <para>
                    This is more of a necessary practice than a "best" practice. When an exception occurs, roll back
                    the <literal>Transaction</literal> and close the <literal>Session</literal>. If you don't, Hibernate
                    can't guarantee that in-memory state accurately represents persistent state. As a special case of this,
                    do not use <literal>Session.load()</literal> to determine if an instance with the given identifier 
                    exists on the database; use <literal>find()</literal> instead. Some exceptions are recoverable, for
                    example the <literal>StaleObjectStateException</literal> and <literal>ObjectNotFoundException</literal>.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Prefer lazy fetching for associations.</term>
            <listitem>
                <para>
                    Use eager (outer-join) fetching sparingly. Use proxies and/or lazy collections for most associations
                    to classes that are not cached at the JVM-level. For associations to cached classes, where there is
                    a high probability of a cache hit, explicitly disable eager fetching using 
                    <literal>outer-join="false"</literal>. When an outer-join fetch is appropriate to a particular use
                    case, use a query with a <literal>left join</literal>.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Consider abstracting your business logic from Hibernate.</term>
            <listitem>
                <para>
                    Hide (Hibernate) data-access code behind an interface. Combine the <emphasis>DAO</emphasis> and 
                    <emphasis>Thread Local Session</emphasis> patterns. You can even have some classes persisted by
                    handcoded JDBC, associated to Hibernate via a <literal>UserType</literal>. (This advice is 
                    intended for "sufficiently large" applications; it is not appropriate for an application with
                    five tables!)
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Implement <literal>equals()</literal> and <literal>hashCode()</literal> using a unique business key.</term>
            <listitem>
                <para>
                    If you compare objects outside of the Session scope, you have to implement <literal>equals()</literal>
                    and <literal>hashCode()</literal>. Inside the Session scope, Java object identity is guaranteed. If
                    you implement these methods, never ever use the database identifier! A transient object doesn't have
                    an identifier value and Hibernate would assign a value when the object is saved. If the object
                    is in a Set while being saved, the hash code changes, breaking the contract. To implement
                    <literal>equals()</literal> and <literal>hashCode()</literal>, use a unique business key, that is,
                    compare a unique combination of class properties. Remember that this key has to be stable and unique
                    only while the object is in a Set, not for the whole lifetime (not as stable as a database primary
                    key). Never use collections in the <literal>equals()</literal> comparison (lazy loading) and be careful
                    with other associated classes that might be proxied.
                </para>
            </listitem>
        </varlistentry>
        <varlistentry>
            <term>Don't use exotic association mappings.</term>
            <listitem>
                <para>
                    Good usecases for a real many-to-many associations are rare. Most of the time you need
                    additional information stored in the "link table". In this case, it is much better to
                    use two one-to-many associations to an intermediate link class. In fact, we think that
                    most associations are one-to-many and many-to-one, you should be careful when using any
                    other association style and ask yourself if it is really neccessary.
                </para>
            </listitem>
        </varlistentry>
    </variablelist>

</chapter>