Available as of Camel 2.0
The asynchronous API in Camel have been rewritten for Camel 2.0, and the information on this page applies for Camel 2.0 and later.
Before we look at these two areas we start with a bit of background information and looks at the concept from at a higher level using diagrams.
The new Async API in Camel 2.0 leverages in much greater detail the Java Concurrency API and its support for executing tasks asynchronous.
When doing messaging there are a few aspects to keep in mind.
First of all a caller can initiate a message exchange as either:
Request only is when the caller sends a message but do not expect any reply. This is also known as fire and forget or event message.
The Request Reply is when the caller sends a message and then waits for a reply. This is like the HTTP protocol that we use every day when we surf the web.
In Camel a message is labeled with a Message Exchange Pattern that labels whether its a request only or request reply message. Camel uses the JBI term for this an uses InOnly for the request only, and InOut for the request reply.
For all message exchange they can be executed either:
A synchronous exchange is defined as the caller sends a message and waits until its complete before continuing. This is illustrated in the diagram below:
1. The client sends a sync Request Reply message over HTTP to Camel. The client application will wait for the response that Camel routes and processes.
On the other hand the asynchronous version is where the caller sends a message to an Endpoint and then returns immediately back to the caller. The message however is processed in another thread, the asynchronous thread. Then the caller can continue doing other work and at the same time the asynchronous thread is processing the message. This is illustrated in the diagram below:
1. The client sends an Async Request Reply message over HTTP to Camel. The control is immediately returned to the client application, that can continue and do other work while Camel routes the message.
You can also do synchronous Request only with Camel. The client sends a message to Camel in which a reply is not expected. However the client still waits until the message is processed completely. This is illustrated in the diagram below:
1. The client sends a Request only and we can still use HTTP despite http being Request Reply by nature.
So why do you want to use synchronous Request Only? Well if you want to know whether the message was processed successfully or not before continuing. With synchronous it allows you to wait while the message is being processed. In case the processing was succesful the control is returned to the client with no notion of error. In case of failure the client can detect this as an exception is thrown. (and exchange.isFailed() returns true).
As opposed to the synchronous Request Only the Async counter part will not wait for the processing of the message to complete. In this case the client can immediately continue doing other work while the message is being routed and processed in Camel. This is illustrated in the diagram below:
1. The client sends a Request only and we can still use HTTP despite http being Request Reply by nature. The control is immediately returned to the client application, that can continue and do other work while Camel routes the message.
Notice: As Camel always returns a Future handle for Async messaging to the client. The client can use this handler to get hold of the status of the processing whether the task is complete or an Exception occured during processing. Note that the client is not required to do so, its perfect valid to just ignore the Future handle.
With these diagrams in mind lets turn out attention to the Async API and how to use it with Camel.
1) The Async Client API
The asyncSend and asyncRequest methods return a Future handle. This handle is what the caller must use later to retrieve the asynchronous response. You can do this by using the extractFutureBody method, or just use plain Java but invoke get() on the Future handle.
The Async Client API with callbacks
These methods also returns the Future handle in case you need them. The difference is that they invokes the callback as well when the Exchange is done being routed.
The java.util.concurrent.Future API have among others the following methods:
Suppose we want to call a HTTP service but it is usually slow and thus we do not want to block and wait for the response, as we can do other important computation. So we can initiate an Async exchange to the HTTP endpoint and then do other stuff while the slow HTTP service is processing our request. And then a bit later we can use the Future handle to get the response from the HTTP service. Yeah nice so lets do it:
First we define some routes in Camel. One for the HTTP service where we simulate a slow server as it takes at least 1 second to reply. And then other route that we want to invoke while the HTTP service is on route. This allows you to be able to process the two routes simultaneously:
And then we have the client API where we call the two routes and we can get the responses from both of them. As the code is based on unit test there is a bit of mock in there as well:
All together it should give you the basic idea how to use this Async API and what it can do.
This example is just to a pure synchronous version of the example from above that was Async based.
The route is the same, so its just how the client initiate and send the messages that differs:
Using the Async API with callbacks
Suppose we want to call a HTTP service but it is usually slow and thus we do not want to block and wait for the response, but instead let a callback gather the response. This allows us to send multiple requests without waiting for the replies before we can send the next request.
First we define a route in Camel for the HTTP service where we simulate a slow server as it takes at least 1 second to reply.
Then we define our callback where we gather the responses. As this is based on an unit test it just gathers the responses in a list. This is a shared callback we use for every request we send in, but you can use your own individual or use an anonymous callback. The callback supports different methods, but we use onDone that is invoked regardless if the Exchange was processed successfully or failed. The org.apache.camel.spi.Synchronization API provides fine grained methods for onCompletion and onFailure for the two situations.
And then we have the client API where we call the HTTP service using asyncCallback 3 times with different input. As the invocation is Async the client will send 3 requests right after each other, so we have 3 concurrent exchanges in progress. The response is gathered by our callback so we do not have to care how to get the response.
Using the Async API with the Camel classic API
When using the Camel API to create a producer and send an Exchange we do it like this:
But to do the same with Async we need a little help from a helper class, so the code is:
In Camel 2.0 the threads DSL replaces the old thread DSL.
The threads DSL leverages the JDK concurrency framework for multi threading. It can be used to turn a synchronous route into Async. What happens is that from the point forwards from threads the messages is routed asynchronous in a new thread. The caller will either wait for a reply if a reply is expected, such as when we use Request Reply messaging. Or the caller will complete as well if no reply was expected such as Request Only messaging.
The threads DSL leverages the JDK concurrency framework for multi threading. It can be used to turn a synchronous route into Async. What happens is that from the point forwards from threads the messages is routed asynchronous in a new thread. Camel leverages the asynchronous routing engine, which was re-introduced in Camel 2.4, to continue routing the Exchange asynchronously.
The threads DSL supports the following options:
The threads DSL uses a thread pool which has a worker queue for tasks. When the worker queue gets full, the task is rejected. You can customize how to react upon this using the rejectedPolicy and callerRunsWhenRejected option. The latter is used for easily switch between the two most common and recommended settings. Either let the current caller thread execute the task (eg it will become synchronous), but also give time for the thread pool to process its current tasks, without adding more tasks - sort of self throttling. This is the default behavior. If setting callerRunsWhenRejected you use the Abort policy, which mean the task is rejected, and a RejectedExecutionException is set on the Exchange, and the Exchange will stop continue being routed, and its UnitOfWork will be regarded as failed.
The other options Discard and DiscardOldest works a bit like Abort, however they do not set any Exception on the Exchange, which mean the Exchange will not be regarded as failed, but the Exchange will be successful. When using Discard and DiscardOldest then the Exchange will not continue being routed. Notice: There is a issue with these two options in Camel 2.9 or below, that cause the UnitOfWork not to be triggered, so we discourage you from using these options in those Camel releases. This has been fixed in Camel 2.10 onwards.
Suppose we receive orders on a JMS queue. Some of the orders expect a reply while other do not (either a JMSReplyTo exists or not). And lets imagine to process this order we need to do some heavy CPU calculation. So how do we avoid the messages that does not expect a reply to block until the entire message is processed? Well we use the threads DSL to turn the route into multi threading asynchronous routing before the heavy CPU task. Then the messages that does not expect a reply can return beforehand. And the messages that expect a reply, well yeah they have to wait anyway. So this can be accomplished like the route below: