JDO : Fetch Groups

When an object is retrieved from the datastore by JDO typically not all fields are retrieved immediately. This is because for efficiency purposes only particular field types are retrieved in the initial access of the object, and then any other objects are retrieved when accessed (lazy loading). The group of fields that are loaded is called a fetch group. There are 3 types of "fetch groups" to consider

  • Default Fetch Group : defined in all JDO specs, containing the fields of a class that will be retrieved by default (with no user specification).
  • Named Fetch Groups : defined by the JDO2 specification, and defined in MetaData (XML/annotations) with the fields of a class that are part of that fetch group. The definition here is static
  • Dynamic Fetch Groups : Programmatic definition of fetch groups at runtime via an API

The fetch group in use for a class is controled via the FetchPlan interface. To get a handle on the current FetchPlan we do

FetchPlan fp = pm.getFetchPlan();

Default Fetch Group

JDO provides an initial fetch group, comprising the fields that will be retrieved when an object is retrieved if the user does nothing to define the required behaviour. By default the default fetch group comprises all fields of the following types :-

  • primitives : boolean, byte, char, double, float, int, long, short
  • Object wrappers of primitives : Boolean, Byte, Character, Double, Float, Integer, Long, Short
  • java.lang.String, java.lang.Number, java.lang.Enum
  • java.math.BigDecimal, java.math.BigInteger
  • java.util.Date

If you wish to change the Default Fetch Group for a class you can update the Meta-Data for the class as follows (for XML)

<class name="MyClass">
    ...
    <field name="fieldX" default-fetch-group="true"/>
</class>

or using annotations

@Persistent(defaultFetchGroup="true")
SomeType fieldX;

When a PersistenceManager is created it starts with a FetchPlan of the "default" fetch group. That is, if we call

Collection fetchGroups = fp.getGroups();

this will have one group, called "default". At runtime, if you have been using other fetch groups and want to revert back to the default fetch group at any time you simply do

fp.setGroup(FetchPlan.DEFAULT);

Named Fetch Groups

As mentioned above, JDO2 allows specification of users own fetch groups. These are specified in the MetaData of the class. For example, if we have the following class

class MyClass
{
    String name;
    HashSet coll;
    MyOtherClass other;
}

and we want to have the other field loaded whenever we load objects of this class, we define our MetaData as

<package name="mydomain">
    <class name="MyClass">
        <field name="name">
            <column length="100" jdbc-type="VARCHAR"/>
        </field>
        <field name="coll" persistence-modifier="persistent">
            <collection element-type="mydomain.Address"/>
            <join/>
        </field>
        <field name="other" persistence-modifier="persistent"/>
        <fetch-group name="otherfield">
            <field name="other"/>
        </fetch-group>
    </class>
</package>

or using annotations

@PersistenceCapable
@FetchGroup(name="otherfield", members={@Persistent(name="other")})
public class MyClass
{
    ...
}

So we have defined a fetch group called "otherfield" that just includes the field with name other. We can then use this at runtime in our persistence code.

PersistenceManager pm = pmf.getPersistenceManager();
pm.getFetchPlan().addGroup("otherfield");

... (load MyClass object)

By default the FetchPlan will include the default fetch group. We have changed this above by adding the fetch group "otherfield", so when we retrieve an object using this PersistenceManager we will be retrieving the fields name AND other since they are both in the current FetchPlan. We can take the above much further than what is shown by defining nested fetch groups in the MetaData. In addition we can change the FetchPlan just before any PersistenceManager operation to control what is fetched during that operation. The user has full flexibility to add many groups to the current Fetch Plan. This gives much power and control over what will be loaded and when. A big improvement over JDO 1.0

The FetchPlan applies not just to calls to PersistenceManager.getObjectById(), but also to PersistenceManager.newQuery(), PersistenceManager.getExtent(), PersistenceManager.detachCopy and much more besides.

To read more about named fetch-groups and how to use it with attach/detach you can look at our Tutorial on DAO Layer design.

Dynamic Fetch Groups

The mechanism above provides static fetch groups defined in XML or annotations. That is great when you know in advance what fields you want to fetch. In some situations you may want to define your fields to fetch at run time. This became standard in JDO2.2 (was previously a DataNucleus extension). It operates as follows

import org.datanucleus.FetchGroup;

// Create a FetchGroup on the PMF called "TestGroup" for MyClass
FetchGroup grp = myPMF.getFetchGroup("TestGroup", MyClass.class);
grp.addMember("field1").addMember("field2");

// Add this group to the fetch plan (using its name)
fp.addGroup("TestGroup");

So we use the DataNucleus PMF as a way of creating a FetchGroup, and then register that FetchGroup with the PMF for use by all PMs. We then enable our FetchGroup for use in the FetchPlan by using its group name (as we do for a static group). The FetchGroup allows you to add/remove the fields necessary so you have full API control over the fields to be fetched.


Fetch Depth

The basic fetch group defines which fields are to be fetched. It doesn't explicitly define how far down an object graph is to be fetched. JDO2 provides two ways of controlling this.

The first is to set the maxFetchDepth for the FetchPlan. This value specifies how far out from the root object the related objects will be fetched. A positive value means that this number of relationships will be traversed from the root object. A value of -1 means that no limit will be placed on the fetching traversal. The default is 1. Let's take an example

public class MyClass1
{
    MyClass2 field1;
    ...
}

public class MyClass2
{
    MyClass3 field2;
    ...
}

public class MyClass3
{
    MyClass4 field3;
    ...
}

and we want to detach field1 of instances of MyClass1, down 2 levels - so detaching the initial "field1" MyClass2 object, and its "field2" MyClass3 instance. So we define our fetch-groups like this

<class name="MyClass1">
    ...
    <fetch-group name="includingField1">
        <field name="field1"/>
    </fetch-group>
</class>
<class name="MyClass2">
    ...
    <fetch-group name="includingField2">
        <field name="field2"/>
    </fetch-group>
</class>

and we then define the maxFetchDepth as 2, like this

pm.getFetchPlan().setMaxFetchDepth(2);

A further refinement to this global fetch depth setting is to control the fetching of recursive fields. This is performed via a MetaData setting "recursion-depth". A value of 1 means that only 1 level of objects will be fetched. A value of -1 means there is no limit on the amount of recursion. The default is 1. Let's take an example

public class Directory
{
    Collection children;
    ...
}
<class name="Directory">
    <field name="children">
        <collection element-type="Directory"/>
    </field>

    <fetch-group name="grandchildren">
        <field name="children" recursion-depth="2"/>
    </fetch-group>
    ...
</class>

So when we fetch a Directory, it will fetch 2 levels of the children field, hence fetching the children and the grandchildren.

Fetch Size

A FetchPlan can also be used for defining the fetching policy when using queries. This can be set using

pm.getFetchPlan().setFetchSize(value);

The default is FetchPlan.FETCH_SIZE_OPTIMAL which leaves it to DataNucleus to optimise the fetching of instances. A positive value defines the number of instances to be fetched. Using FetchPlan.FETCH_SIZE_GREEDY means that all instances will be fetched immediately.