spring-framework/spring-framework-reference/src/scheduling.xml

<?xml version="1.0" encoding="UTF-8" ?>
<chapter id="scheduling">
    <title>Scheduling and Thread Pooling</title>
	<section id="scheduling-introduction">
		<title>Introduction</title>
		<para>
		The Spring Framework features integration classes for scheduling support. Currently, Spring
		supports the <classname>Timer</classname>, part of the JDK since 1.3, and the
		Quartz Scheduler (<ulink url="http://www.opensymphony.com/quartz/"/>). Both schedulers
		are set up using a <interfacename>FactoryBean</interfacename> with optional references
		to <classname>Timer</classname> or <classname>Trigger</classname> instances, respectively.
		Furthermore, a convenience class for both the Quartz Scheduler and the <classname>Timer</classname> is
		available that allows you to invoke a method of an existing target object
		(analogous to the normal <classname>MethodInvokingFactoryBean</classname> operation).
		Spring also features classes for thread pooling that abstract
		away differences between Java SE 1.4, Java SE 5 and Java EE environments.
  	</para>
	</section>
	<section id="scheduling-quartz">
		<title>Using the OpenSymphony Quartz Scheduler</title>
		<para>Quartz uses <classname>Trigger</classname>, <classname>Job</classname> and 
		<classname>JobDetail</classname> objects to realize scheduling of all kinds of jobs.
		For the basic concepts behind Quartz, have a look at
		<ulink url="http://www.opensymphony.com/quartz" />. For convenience purposes,
		Spring offers a couple of classes that simplify the usage of Quartz within
		Spring-based applications.
		</para>
		<section id="scheduling-quartz-jobdetail">
			<title>Using the JobDetailBean</title>
			<para>
			<classname>JobDetail</classname> objects contain all information needed to
			run a job. The Spring Framework provides a <classname>JobDetailBean</classname>
			that makes the <classname>JobDetail</classname> more of an actual JavaBean
			with sensible defaults. Let's have a look at an example:
			</para>
			<programlisting language="xml"><![CDATA[
<bean name="exampleJob" class="org.springframework.scheduling.quartz.JobDetailBean">
  <property name="jobClass" value="example.ExampleJob" />
  <property name="jobDataAsMap">
    <map>
      <entry key="timeout" value="5" />
    </map>
  </property>
</bean>]]></programlisting>
			<para>The job detail bean has all information it needs to run the job (<classname>ExampleJob</classname>).
			The timeout is specified in the job data map. The job data map is
			available through the <classname>JobExecutionContext</classname>
			(passed to you at execution time), but the <classname>JobDetailBean</classname>
			also maps the properties from the job data map to properties of the actual job.
			So in this case, if the <classname>ExampleJob</classname> contains a property
			named <literal>timeout</literal>, the <classname>JobDetailBean</classname> will
			automatically apply it:</para>
			<programlisting language="java"><![CDATA[package example;

public class ExampleJob extends QuartzJobBean {

  private int timeout;
  
  /**
   * Setter called after the ExampleJob is instantiated
   * with the value from the JobDetailBean (5)
   */ 
  public void setTimeout(int timeout) {
    this.timeout = timeout;
  }
  
  protected void executeInternal(JobExecutionContext ctx) throws JobExecutionException {
      ]]><lineannotation>// do the actual work</lineannotation><![CDATA[
  }
}]]></programlisting>
			<para>All additional settings from the job detail bean are of course available to you as well.</para>
			<para><emphasis>Note: Using the <literal>name</literal> and <literal>group</literal> properties,
			you can modify the name and the group of the job, respectively. By default, the name of
			the job matches the bean name of the job detail bean (in the example above, this is
			<literal>exampleJob</literal>).</emphasis></para>
		</section>
		<section id="scheduling-quartz-method-invoking-job">
			<title>Using the <classname>MethodInvokingJobDetailFactoryBean</classname></title>
			<para>Often you just need to invoke a method on a specific object. Using the
			<classname>MethodInvokingJobDetailFactoryBean</classname> you can do exactly this:</para>
			<programlisting language="xml"><![CDATA[<bean id="jobDetail" class="org.springframework.scheduling.quartz.MethodInvokingJobDetailFactoryBean">
  <property name="targetObject" ref="exampleBusinessObject" />
  <property name="targetMethod" value="doIt" />
</bean>]]></programlisting>
			<para>The above example will result in the <literal>doIt</literal> method being called on the
			<literal>exampleBusinessObject</literal> method (see below):</para>
			<programlisting language="java"><![CDATA[public class ExampleBusinessObject {
  
  ]]><lineannotation>// properties and collaborators</lineannotation><![CDATA[
  
  public void doIt() {
    ]]><lineannotation>// do the actual work</lineannotation><![CDATA[
  }
}]]></programlisting>

			<programlisting language="xml"><![CDATA[
<bean id="exampleBusinessObject" class="examples.ExampleBusinessObject"/>]]></programlisting>
			<para>Using the <classname>MethodInvokingJobDetailFactoryBean</classname>, you don't need to
			create one-line jobs that just invoke a method, and you only need to create the actual
			business object and wire up the detail object.</para>
			<para>By default, Quartz Jobs are stateless, resulting in the possibility of jobs interfering
			with each other. If you specify two triggers for the same <classname>JobDetail</classname>,
			it might be possible that before the first job has finished, the second one will start.
			If <classname>JobDetail</classname> classes implement the
			<interfacename>Stateful</interfacename> interface, this won't happen. The second job
			will not start before the first one has finished. To make jobs resulting from the
			<classname>MethodInvokingJobDetailFactoryBean</classname> non-concurrent, set the
			<literal>concurrent</literal> flag to <literal>false</literal>.</para>
			<programlisting language="xml"><![CDATA[
<bean id="jobDetail" class="org.springframework.scheduling.quartz.MethodInvokingJobDetailFactoryBean">
  <property name="targetObject" ref="exampleBusinessObject" />
  <property name="targetMethod" value="doIt" />
  <property name="concurrent" value="false" />
</bean>]]></programlisting>
			<note>
				<para>By default, jobs will run in a concurrent fashion.</para>
			</note>
		</section>
		<section id="scheduling-quartz-cron">
			<title>Wiring up jobs using triggers and the <classname>SchedulerFactoryBean</classname></title>
			<para>
				We've created job details and jobs. We've also reviewed the convenience bean
				that allows to you invoke a method on a specific object. Of course, we still need
				to schedule the jobs themselves. This is done using triggers and a 
				<classname>SchedulerFactoryBean</classname>. Several triggers are available
				within Quartz. Spring offers two subclassed triggers with convenient defaults:
				<classname>CronTriggerBean</classname> and <classname>SimpleTriggerBean</classname>.
			</para>
			<para>
				Triggers need to be scheduled. Spring offers a <classname>SchedulerFactoryBean</classname>
				that exposes triggers to be set as properties. <classname>SchedulerFactoryBean</classname>
				schedules the actual jobs with those triggers.
			</para>
			<para>Find below a couple of examples:</para>
			<programlisting language="xml"><![CDATA[<bean id="simpleTrigger" class="org.springframework.scheduling.quartz.SimpleTriggerBean">
    <!-- see the example of method invoking job above -->
    <property name="jobDetail" ref="jobDetail" />
    <!-- 10 seconds -->
    <property name="startDelay" value="10000" />
    <!-- repeat every 50 seconds -->
    <property name="repeatInterval" value="50000" />
</bean>

<bean id="cronTrigger" class="org.springframework.scheduling.quartz.CronTriggerBean">
    <property name="jobDetail" ref="exampleJob" />
    <!-- run every morning at 6 AM -->
    <property name="cronExpression" value="0 0 6 * * ?" />
</bean>]]></programlisting>
			<para>Now we've set up two triggers, one running every 50 seconds with a starting delay of
			10 seconds and one every morning at 6 AM. To finalize everything, we need to set up the
			<classname>SchedulerFactoryBean</classname>:</para>
			<programlisting language="xml"><![CDATA[<bean class="org.springframework.scheduling.quartz.SchedulerFactoryBean">
    <property name="triggers">
        <list>
            <ref bean="cronTrigger" />
            <ref bean="simpleTrigger" />
        </list>
    </property>
</bean>]]></programlisting>
			<para>
				More properties are available for the <classname>SchedulerFactoryBean</classname> for you
				to set, such as the	calendars used by the job details, properties to customize Quartz with,
				etc. Have a look at the
				<ulink url="http://static.springframework.org/spring/docs/2.5.x/api/org/springframework/scheduling/quartz/SchedulerFactoryBean.html">SchedulerFactoryBean Javadoc</ulink>
				for more information.
			</para>
		</section>
	</section>
	<section id="scheduling-jdk-timer">
		<title>Using JDK Timer support</title>
		<para>
  			The other way to schedule jobs in Spring is to use JDK
			<classname>Timer</classname> objects. You can create custom timers or
			use the timer that invokes methods. Wiring timers is done using the
			<classname>TimerFactoryBean</classname>.
  		</para>
		<section id="scheduling-jdk-timer-creating">
			<title>Creating custom timers</title>
			<para>
  				Using the <classname>TimerTask</classname> you can create customer
				timer tasks, similar to Quartz jobs:
			</para>
  			<programlisting language="java"><![CDATA[public class CheckEmailAddresses extends TimerTask {

  private List emailAddresses;
  
  public void setEmailAddresses(List emailAddresses) {
    this.emailAddresses = emailAddresses;
  }
  
  public void run() {
    ]]><lineannotation>// iterate over all email addresses and archive them</lineannotation><![CDATA[
  }
}]]></programlisting>
			<para>
				Wiring it up is simple:
			</para>
			<programlisting language="xml"><![CDATA[<bean id="checkEmail" class="examples.CheckEmailAddress">
    <property name="emailAddresses">
        <list>
            <value>test@springframework.org</value>
            <value>foo@bar.com</value>
            <value>john@doe.net</value>
        </list>
    </property>
</bean>

<bean id="scheduledTask" class="org.springframework.scheduling.timer.ScheduledTimerTask">
    ]]><lineannotation>&lt;!-- wait 10 seconds before starting repeated execution --&gt;</lineannotation><![CDATA[
    <property name="delay" value="10000" />
    ]]><lineannotation>&lt;!-- run every 50 seconds --&gt;</lineannotation><![CDATA[
    <property name="period" value="50000" />
    <property name="timerTask" ref="checkEmail" />
</bean>]]></programlisting>
			<para>
				<emphasis>
					Note that letting the task only run once can be done by changing the
					<literal>period</literal> property to 0 (or a negative value).
				</emphasis>
			</para>
		</section>
		<section id="scheduling-jdk-timer-method-invoking-task">
			<title>Using the <classname>MethodInvokingTimerTaskFactoryBean</classname></title>
			<para>
				Similar to the Quartz support, the <classname>Timer</classname> support also features
				a component that allows you to periodically invoke a method:
			</para>
			<programlisting language="xml"><![CDATA[<bean id="doIt" class="org.springframework.scheduling.timer.MethodInvokingTimerTaskFactoryBean">
    <property name="targetObject" ref="exampleBusinessObject" />
    <property name="targetMethod" value="doIt" />
</bean>]]></programlisting>
			<para>
				The above example will result in the <literal>doIt</literal> method being called on the
				<literal>exampleBusinessObject</literal> (see below):
			</para>
			<programlisting language="java"><![CDATA[public class BusinessObject {
  
  ]]><lineannotation>// properties and collaborators</lineannotation><![CDATA[
  
  public void doIt() {
    ]]><lineannotation>// do the actual work</lineannotation><![CDATA[
  }
}]]></programlisting>
			<para>Changing the <literal>timerTask</literal> reference of the
			<classname>ScheduledTimerTask</classname> example to the bean <literal>doIt</literal>
			will result in the <literal>doIt</literal> method being executed on a fixed schedule.</para>
		</section>
		<section id="scheduling-jdk-timer-factory-bean">
			<title>Wrapping up: setting up the tasks using the <classname>TimerFactoryBean</classname></title>
			<para>The <classname>TimerFactoryBean</classname> is similar to the Quartz
			<classname>SchedulerFactoryBean</classname> in that it serves the same
			purpose: setting up the actual scheduling. The <classname>TimerFactoryBean</classname>
			sets up an actual <classname>Timer</classname> and schedules the tasks it has
			references to. You can specify whether or not daemon threads should be used.</para>
			<programlisting language="xml"><![CDATA[<bean id="timerFactory" class="org.springframework.scheduling.timer.TimerFactoryBean">
    <property name="scheduledTimerTasks">
        <list>
            ]]><lineannotation>&lt;!-- see the example above --&gt;</lineannotation><![CDATA[
            <ref bean="scheduledTask" />
        </list>
    </property>
</bean>]]></programlisting>
		</section>
	</section>
	<section id="scheduling-task-executor">
		<title>The Spring <interfacename>TaskExecutor</interfacename> abstraction</title>
		<para>Spring 2.0 introduces a new abstraction for dealing with
		executors. Executors are the Java 5 name for the concept of
		thread pools. The "executor" naming is due to the fact that there
		is no guarantee that the underlying implementation is actually a
		pool; an executor may be single-threaded or even synchronous.
		Spring's abstraction hides implementation details between
		Java SE 1.4, Java SE 5 and Java EE environments.</para>
		<title>The <interfacename>TaskExecutor</interfacename> interface</title>
		<para>Spring's <interfacename>TaskExecutor</interfacename> interface is
		identical to the <classname>java.util.concurrent.Executor</classname>
		interface. In fact, its primary reason for existence is to abstract away
		the need for Java 5 when using thread pools. The interface has a single
		method <classname>execute(Runnable task)</classname> that accepts a task
		for execution based on the semantics and configuration of the thread pool.</para>
		<para>The <interfacename>TaskExecutor</interfacename> was originally
		created to give other Spring components an abstraction for thread pooling where
		needed. Components such as the <classname>ApplicationEventMulticaster</classname>,
		JMS's <classname>AbstractMessageListenerContainer</classname>,
		and Quartz integration all use the <interfacename>TaskExecutor</interfacename>
		abstraction to pool threads. However, if your beans need thread pooling behavior,
		it is possible to use this abstraction for your own needs.</para>
		
		<section id="scheduling-task-executor-types">
			<title><interfacename>TaskExecutor</interfacename> types</title>
			<para>There are a number of pre-built implementations of
			<interfacename>TaskExecutor</interfacename> included with the
			Spring distribution. In all likelihood, you shouldn't ever
			need to implement your own.</para>
			
			<itemizedlist>
				<listitem>
					<para>
						<classname>SimpleAsyncTaskExecutor</classname>
					</para>

					<para>
						This implementation does not reuse any threads,
						rather it starts up a new thread for each
						invocation. However, it does support a
						concurrency limit which will block any
						invocations that are over the limit until a slot
						has been freed up. If you're looking for true
						pooling, keep scrolling further down the page.
					</para>
				</listitem>
				
				<listitem id="syncTaskExecutor">
					<para>
						<classname>SyncTaskExecutor</classname>
					</para>

					<para>
						This implementation doesn't execute
						invocations asynchronously. Instead, each
						invocation takes place in the calling thread. It
						is primarily used in situations where
						mutlithreading isn't necessary such as simple
						test cases.
					</para>
				</listitem>
				
				<listitem id="concurrentTaskExecutor">
					<para>
						<classname>ConcurrentTaskExecutor</classname>
					</para>

					<para>
						This implementation is a wrapper for a Java 5
						<classname>java.util.concurrent.Executor</classname>.
						There is an alternative,
						<classname>ThreadPoolTaskExecutor</classname>,
						that exposes the <classname>Executor</classname>
						configuration parameters as bean properties. It
						is rare to need to use the <classname>ConcurrentTaskExecutor</classname>
						but if the
						<link linkend="threadPoolTaskExecutor"><classname>ThreadPoolTaskExecutor</classname></link>
						isn't robust enough for your needs, the
						<classname>ConcurrentTaskExecutor</classname>
						is an alternative.
					</para>
				</listitem>
				
				<listitem id="simpleThreadPoolTaskExecutor">
					<para>
						<classname>SimpleThreadPoolTaskExecutor</classname>
					</para>
					
					<para>
						This implementation is actually a subclass of
						Quartz's <classname>SimpleThreadPool</classname>
						which listens to Spring's lifecycle callbacks.
						This is typically used when you have a
						threadpool that may need to be shared by both
						Quartz and non-Quartz components.
					</para>
				</listitem>

				<listitem id="threadPoolTaskExecutor">
					<para>
						<classname>ThreadPoolTaskExecutor</classname>
					</para>
					
					<sidebar>
						<para>
							It is not possible to use any backport or
							alternate versions of the
							<classname>java.util.concurrent</classname>
							package with this implementation. Both Doug
							Lea's and Dawid Kurzyniec's implementations
							use different package structures which will
							prevent them from working correctly.
						</para>
					</sidebar>

					<para>
						This implementation can only be used in a Java 5
						environment but is also the most commonly used
						one in that environment. It exposes bean properties for
						configuring a 
						<classname>java.util.concurrent.ThreadPoolExecutor</classname>
						and wraps it in a <interfacename>TaskExecutor</interfacename>.
						If you need something advanced such as a
						<classname>ScheduledThreadPoolExecutor</classname>,
						it is recommended that you use a
						<link linkend="concurrentTaskExecutor"><classname>ConcurrentTaskExecutor</classname></link>
						instead.
                    </para>
                </listitem>
				
				<listitem>
					<para>
						<classname>TimerTaskExecutor</classname>
					</para>

					<para>
						This implementation uses a single
						<classname>TimerTask</classname>
						as its backing implementation. It's different
						from the
						<link linkend="syncTaskExecutor"><classname>SyncTaskExecutor</classname></link>
						in that the method invocations are executed in a
						separate thread, although they are synchronous
						in that thread.
					</para>
                </listitem>
				
				<listitem>
					<para>
						<classname>WorkManagerTaskExecutor</classname>
					</para>
					
					<sidebar><para>
							CommonJ is a set of specifications jointly
							developed between BEA and IBM. These
							specifications are not Java EE standards, but
							are standard across BEA's and IBM's
							Application Server implementations.
						</para></sidebar>

					<para>
						This implementation uses the CommonJ WorkManager
						as its backing implementation and is the central
						convenience class for setting up a CommonJ
						WorkManager reference in a Spring context.
						Similar to the
						<link linkend="simpleThreadPoolTaskExecutor"><classname>SimpleThreadPoolTaskExecutor</classname></link>,
						this class implements the WorkManager
						interface and therefore can be used directly as
						a WorkManager as well.
					</para>
                </listitem>

			</itemizedlist>
		</section>
		<section id="scheduling-task-executor-usage">
			<title>Using a <interfacename>TaskExecutor</interfacename></title>
			<para>Spring's <interfacename>TaskExecutor</interfacename> implementations
			are used as simple JavaBeans.  In the example below, we define
			a bean that uses the <classname>ThreadPoolTaskExecutor</classname>
			to asynchronously print out a set of messages.</para>

			<programlisting language="java"><![CDATA[import org.springframework.core.task.TaskExecutor;

public class TaskExecutorExample {

  private class MessagePrinterTask implements Runnable {

    private String message;

    public MessagePrinterTask(String message) {
      this.message = message;
    }

    public void run() {
      System.out.println(message);
    }

  }

  private TaskExecutor taskExecutor;

  public TaskExecutorExample(TaskExecutor taskExecutor) {
    this.taskExecutor = taskExecutor;
  }

  public void printMessages() {
    for(int i = 0; i < 25; i++) {
      taskExecutor.execute(new MessagePrinterTask("Message" + i));
    }
  }
}]]></programlisting>
			
			<para>As you can see, rather than retrieving a thread from the
			pool and executing yourself, you add your <classname>Runnable</classname>
			to the queue and the <interfacename>TaskExecutor</interfacename>
			uses its internal rules to decide when the task gets executed.</para>
			
			<para>To configure the rules that the <interfacename>TaskExecutor</interfacename>
			will use, simple bean properties have been exposed.</para>
			
			<programlisting language="xml"><![CDATA[<bean id="taskExecutor" class="org.springframework.scheduling.concurrent.ThreadPoolTaskExecutor">
  <property name="corePoolSize" value="5" />
  <property name="maxPoolSize" value="10" />
  <property name="queueCapacity" value="25" />
</bean>

<bean id="taskExecutorExample" class="TaskExecutorExample">
  <constructor-arg ref="taskExecutor" />
</bean>]]></programlisting>

		</section>
	</section>
</chapter>