Scala + Akka + ning/async-http-client - Vancouver Scala meetup February 2015
Download as PPTX, PDF13 likes2,622 views
The document discusses the integration of Scala, Akka, and Ning's async-http-client to optimize network latency and improve webfeed performance. It highlights the transition from a resource-heavy architecture to a more efficient actor-based model, resulting in increased throughput and reduced CPU utilization. Key lessons emphasize the importance of managing back pressure, simplifying actor responsibilities, and adapting the architecture as needed.
Scala + Akka + ning/async-http-client - Vancouver Scala meetup February 2015
- 2. Scala + Akka + ning/async-http-client Or: How I stopped worrying about network latency and drank the reactive cool-aid
- 4. RSS is dead, long live RSS! ● In constant decline over the years as consumer-facing tech ● Still very popular between “content services” ● RSS/Atom feeds @ HS o Follow feeds on behalf of users o Post to social networks based on content o ~250K active feeds, ~400K messages daily
- 5. Scaling webfeeds using PHP
- 6. But latency isn’t the problem; blocking is.
- 7. Enter Scala + Akka + async HTTP client
- 8. Akka: a short detour for the uninitiated “Toolkit and runtime for building highly concurrent, distributed, and fault tolerant applications on the JVM” ● Implementation of the Actor model ● Scala, Java ● Adheres strictly to the Reactive Manifesto: o Responsive, Resilient, Elastic, Message-driven
- 11. … in an actor
- 12. Actors in Scala webfeed service
- 13. So, what did this buy us?
- 14. Now BeforeBefore ● 120 processes ● 6 X m3.xlarge ● 2 processes ● 2 X m3.xlarge
- 15. 90 webfeed/sec 300 webfeed/sec
- 16. CPU utilization & load average - barely breaking a sweat!
- 17. Some lessons learned ● You can't ignore back pressure o Pull-based approach o Akka Streams to the rescue? ● scala.xml.pull.XMLEventReader is not production ready ● Be weary of the resulting load on the rest of your infrastructure!
- 18. Some lessons learned (continued) ● Actors should only ever do one thing (and do it well) o Easy to modify the actor topology (and it will happen!) o Easier to describe supervisor’s responsibilities ● Let it crash!
Editor's Notes
- #3: TODO: add owly loves scala pic
- #4: -Who I am -What does Hootsuite do -What the Publishing team does -Mention move to service-oriented/microservices architecture -Scala+Akka at Hootsuite (not sure I need some slides for this material)
- #5: -So, what am I going to talk about? RSS/Atom feeds? -XML? Is this guy for real? -Constant decline over the years as consumer-facing technology (graph) -But still used a great deal between "content services" -How are RSS/Atom feeds used at Hootsuite? -Webfeeds at Hootsuite; how much?
- #6: -Show a rough sketch of php+gearman+cron -Latency, latency, latency everywhere!
- #7: Talk about the example of mailing a letter or getting your car/bike fixed Waiting around for something that someone else is doing to be finished is simply wasteful.
- #10: actors (encapsulate state and behaviour - true encapsulation) mailboxes immutable messages supervisory hierarchy concurrency and distribution is handled by the framework
- #11: Ning’s async-http-client -library written in Java but some Scala wrappers exist -we use Play’s wrapper for convenience but access the underlying ning client for greater control of the configuration -easy to configure -returns a Future to map, compose, etc. (a familiar construct to Scala hackers)
- #12: Common pattern for handling futures in Akka actors.
- #15: -Graphs show rough representation of throughput in # of message sent by units of time -Greater throughput with less computational capacity
- #16: -Number of webfeeds processed per second. -This is after throttling the service because of the pressure it imposed on other parts of the system