About Tomasz Nurkiewicz

Java EE developer, Scala enthusiast. Enjoying data analysis and visualization. Strongly believes in the power of testing and automation.

Non-blocking I/O – discovering Akka

Here comes the time to follow some good practices when implementing actors. One of the most important rules we should follow is avoiding any blocking input/output operations, polling, busy waiting, sleeping, etc. Simply put, actor while handling a message should only depend on CPU and if it doesn’t need CPU cycles it should immediately return from receive and let other actors to process. If we follow this rule strictly, Akka can easily handle hundreds of thousands of messages per second using just a handful of threads. It shouldn’t come as a surprise that even though our application can comprise thousands of seemingly independent actors (e.g. one actor per each HTTP connection, one player in MMO game, etc.), each actor gets only a limited CPU time within a pool of threads. With default 10 threads handling all the actors in the system, one blocking or sleeping actor is enough to reduce the throughput by 10%. Therefore 10 actors sleeping at the same time completely halt the system.

For that reason calling sleep() or actively waiting for a response from some other actor is highly discouraged within receive. Unfortunately there is no mature asynchronous library equivalent to JDBC (but watch postgresql-netty, adbcj, async-mysql-connector and also related: mongodb-async-driver) and using NIO is rather problematic. But we should seek for alternatives and avoid blocking code whenever possible. In our sample application fetching random numbers from external web service was implemented in quite naive way:

val url = new URL("https://www.random.org/integers/?num=" + batchSize + "&min=0&max=65535&col=1&base=10&format=plain&rnd=new")
val connection = url.openConnection()
val stream = Source.fromInputStream(connection.getInputStream)
sender ! RandomOrgServerResponse(stream.getLines().map(_.toInt).toList)
stream.close()

This code blocks waiting for an HTTP reply for up to one minute. This means our actor can’t handle any other message for several seconds. Moreover it holds one thread (by default out of ten) from Akka worker pool. This pool is suppose to be shared among thousands of actors, so that feels a bit selfish.

Luckily mature asynchronous HTTP client libraries exist, namely async-http-client (based on netty, with Scala wrapper called Dispatch) and Jetty HttpClient. For test purposes we’ll use the first one (and leave Dispatch for later). API is quite obvious: it asks us for target URL and a callback object, which will be used when reply arrives. Thus sending HTTP request is asynchronous and non-blocking (actor can quickly consume more incoming messages) and the response arrives asynchronously from a different thread:

implicit def block2completionHandler[T](block: Response => T) = new AsyncCompletionHandler[T]() {
  def onCompleted(response: Response) = block(response)
}

def receive = {
  case FetchFromRandomOrg(batchSize) =>
    val curSender = sender
    val url = "https://www.random.org/integers/?num=" + batchSize + "&min=0&max=65535&col=1&base=10&format=plain&rnd=new"
    client.prepareGet(url).execute {
      response: Response =>
        val numbers = response.getResponseBody.lines.map(_.toInt).toList
        curSender ! RandomOrgServerResponse(numbers)
    }
}

We are very close to really dangerous bug in the code above. Notice how I make a local copy of sender called curSender. If I wouldn’t do this, the block of code executed when a response arrives would read current value of sender. Current, that is if our actor was handling some other message at a time, it would point to a sender of that other message. As a side note that’s one of the reasons why variables accessed from anonymous inner classes in Java have to be final. It’s also a good reason to avoid arbitrary code blocks called asynchronously inside actors. It’s much better to extract them to a separate class outside of the actor to avoid accidental access to internal actor state.

Let’s leave our example for a while. Imagine how scalable our general architecture would be for a general RSS/Atom feed reader as a service. For each feed URL we can create one actor (and we monitor thousands of feeds, thus that many actors). Actor sends an asynchronous request to each site and waits for a response. Theoretically using just one worker thread we can handle thousands of feeds/servers, processing the results on the fly as they come (after all each server has a different response time). In a classic, blocking model we can only process as many feeds concurrently as many threads we can use (certainly not several thousands), not to mention each thread requires significant amount of memory.

If you see some similarities to node.js you are on the right track. This framework is based entirely on asynchronous I/O, thus being able to handle large amount of concurrent connections using only one (!) thread.

Source code for this article is available on GitHub in non-blocking-io tag.

This was a translation of my article “Poznajemy Akka: nieblokujace I/O” originally published on scala.net.pl.
 

Reference: Non-blocking I/O – discovering Akka from our JCG partner Tomasz Nurkiewicz at the Java and neighbourhood blog.

Do you want to know how to develop your skillset to become a Java Rockstar?

Subscribe to our newsletter to start Rocking right now!

To get you started we give you two of our best selling eBooks for FREE!

JPA Mini Book

Learn how to leverage the power of JPA in order to create robust and flexible Java applications. With this Mini Book, you will get introduced to JPA and smoothly transition to more advanced concepts.

JVM Troubleshooting Guide

The Java virtual machine is really the foundation of any Java EE platform. Learn how to master it with this advanced guide!

Given email address is already subscribed, thank you!
Oops. Something went wrong. Please try again later.
Please provide a valid email address.
Thank you, your sign-up request was successful! Please check your e-mail inbox.
Please complete the CAPTCHA.
Please fill in the required fields.

Leave a Reply


five × 8 =



Java Code Geeks and all content copyright © 2010-2014, Exelixis Media Ltd | Terms of Use | Privacy Policy | Contact
All trademarks and registered trademarks appearing on Java Code Geeks are the property of their respective owners.
Java is a trademark or registered trademark of Oracle Corporation in the United States and other countries.
Java Code Geeks is not connected to Oracle Corporation and is not sponsored by Oracle Corporation.
Do you want to know how to develop your skillset and become a ...
Java Rockstar?

Subscribe to our newsletter to start Rocking right now!

To get you started we give you two of our best selling eBooks for FREE!

Get ready to Rock!
You can download the complementary eBooks using the links below:
Close