Difference between Spring MVC's @Async, Deferr

Question

I've a long-running task defined in a Spring service. It is started by a Spring MVC controller. I want to start the service and return back an HttpResponse to the caller before the service ends. The service saves a file on file system at end. In javascript I've created a polling job to check service status.

In Spring 3.2 I've found the @Async annotation, but I don't understand how it is different from DeferredResult and Callable. When do I have to use @Async and when should I use DeferredResult?

Answer 1

Your controller is eventually a function executed by the servlet container (I will assume it is Tomcat) worker thread. Your service flow start with Tomcat and ends with Tomcat. Tomcat gets the request from the client, holds the connection, and eventually returns a response to the client. Your code (controller or servlet) is somewhere in the middle.

Consider this flow:

1. Tomcat get client request.
2. Tomcat executes your controller.
3. Release Tomcat thread but keep the client connection (don't return response) and run heavy processing on different thread.
4. When your heavy processing complete, update Tomcat with its response and return it to the client (by Tomcat).

Because the servlet (your code) and the servlet container (Tomcat) are different entities, then to allow this flow (releasing tomcat thread but keep the client connection) we need to have this support in their contract, the package javax.servlet, which introduced in Servlet 3.0 . Now, getting back to your question, Spring MVC use the new Servlet 3.0 capability when the return value of the controller is DeferredResult or Callable, although they are two different things. Callable is an interface that is part of java.util, and it is an improvement for the Runnable interface (should be implemented by any class whose instances are intended to be executed by a thread). Callable allows to return a value, while Runnable does not. DeferredResult is a class designed by Spring to allow more options (that I will describe) for asynchronous request processing in Spring MVC, and this class just holds the result (as implied by its name) while your Callable implementation holds the async code. So it means you can use both in your controller, run your async code with Callable and set the result in DeferredResult, which will be the controller return value. So what do you get by using DeferredResult as the return value instead of Callable? DeferredResult has built-in callbacks like onError, onTimeout, and onCompletion. It makes error handling very easy.In addition, as it is just the result container, you can choose any thread (or thread pool) to run on your async code. With Callable, you don't have this choice.

Regarding @Async, it is much more simple – annotating a method of a bean with @Async will make it execute in a separate thread. By default (can be overridden), Spring uses a SimpleAsyncTaskExecutor to actually run these methods asynchronously.

In conclusion, if you want to release Tomcat thread and keep the connection with the client while you do heavy processing, then your controller should return Callable or DeferredResult. Otherwise, you can run the code on method annotated with @Async.

Answer 2

DeferredResult takes advantage of the Servlet 3.0 AsyncContext. It will not block the thread like the others will when you need a result returned.

Another big benefit is that DeferredResult supports callbacks.

Answer 3

Async annotates a method so it is going to be called asynchronously.

@org.springframework.stereotype.Service
public class MyService {
    @org.springframework.scheduling.annotation.Async
    void DoSomeWork(String url) {
        [...]
    }
}

So Spring could do so you need to define how is going to be executed. For example:

<task:annotation-driven />
<task:executor id="executor" pool-size="5-10" queue-capacity="100"/>

This way when you call service.DoSomeWork("parameter") the call is put into the queue of the executor to be called asynchronously. This is useful for tasks that could be executed concurrently.

You could use Async to execute any kind of asynchronous task. If what you want is calling a task periodically you could use @Scheduled (and use task:scheduler instead of task:executor). They are simplified ways of calling java Runnables.

DeferredResult<> is used to answer to a petition without blocking the Tomcat HTTP thread used to answer. Usually is going to be the return value for a ResponseBody annotated method.

@org.springframework.stereotype.Controller
{
    private final java.util.concurrent.LinkedBlockingQueue<DeferredResult<String>> suspendedRequests = new java.util.concurrent.LinkedBlockingQueue<>();

    @RequestMapping(value = "/getValue")
    @ResponseBody
    DeferredResult<String> getValue() {
            final DeferredResult<String> result = new DeferredResult<>(null, null);
            this.suspendedRequests.add(result);
            result.onCompletion(new Runnable() {
            @Override
            public void run() {
        suspendedRequests.remove(result);
            }
});
            service.setValue(result); // Sets the value!
            return result;
    }
}

The previous example lacks one important thing and it's that doesn't show how the deferred result is going to be set. In some other method (probably the setValue method) there is going to be a result.setResult(value). After the call to setResult Spring is going to call the onCompletion procedure and return the answer to the HTTP request (see https://en.wikipedia.org/wiki/Push_technology#Long_polling).

But if you just are executing the setValue synchronously there is no advantage in using a deferred result.Here is where Async comes in hand. You could use an async method to set the return value in some point in the future using another thread.

    @org.springframework.scheduling.annotation.Async
    void SetValue(DeferredResult<String> result) {
        String value;
        // Do some time consuming actions
        [...]
        result.setResult(value);
    }

Async is not needed to use a deferred result, its just one way of doing it.

In the example there is a queue of deferred results that, for example, a scheduled task could be monitoring to process it's pending requests. Also you could use some non blocking mechanism (see http://en.wikipedia.org/wiki/New_I/O) to set the returning value.

To complete the picture you could search information about java standard futures (http://docs.oracle.com/javase/1.5.0/docs/api/java/util/concurrent/Future.html) and callables (http://docs.oracle.com/javase/1.5.0/docs/api/java/util/concurrent/Callable.html) that are somewhat equivalent to Spring DeferredResult and Async.