an introduction to reactive applications, reactive streams, and options for the jvm (sacon sf 2016)

An introduction to reactive applications, Reactive Streams, and

options for the JVMSteve Pember

CTO, ThirdChannel

Software Architecture Con, 2016

THIRDCHANNEL @svpember


Agenda• The Problem

• To Be Reactive

• What are Reactive Streams?

• Rx in Depth

• An Overview of JVM Options

• Demo Time!


The Problem: The Need to go Reactive

Really, it’s Two problems


1) Performance Demands Are Always Increasing

We Use Technology from the Beginning of Web Development

Things Slow Down

Users get angry quickly

–Johnny Appleseed

“Type a quote here.”

Let’s Keep Our Users Happy And Engaged


2) The Rise of Microservices

Multiple Integration Points

It’s Not Only Users That Use Up Resources

So… what to do?

Embrace Reactive Applications


Reactive Applications• Responsive



• Resilient

Embrace Failure

Independent Things Fail Independently



• Resilient

• Elastic (Scalable)



• Resilient

• Elastic (Scalable)

• Asynchronous / Message Driven

Free up resources with Async Operations & Non-Blocking I/O

Now do all of this at every level of your app…

All the Way Down


• User Interface: What the user interacts with (JS, iOS, Android)

• Inter-Service: Distributed application topology + service communication design (e.g. Kafka for async messaging)

• Intra-Service: The code! (well designed with bounded contexts)

• Framework: Structures your code + provides external communication, DI, etc

• Execution Environment: Broad area, but includes how your code is executed (e.g. in JVM: Tomcat, Jetty, Netty)


Mostly talk about this…

We’ll talk a little about this…

And a little more about this

There’s not enough time!

Writing Reactive Code is the best method of introduction…

Async is Hard for Humans

One Excellent Tool is (are?) Reactive Streams

But Wait…

“Reactive Streams”, “Reactive Extensions”, or “Rx”

Collections + Time

Single abstraction over data from many sources

Observer Pattern Push (not Pull) based Iterators

Stream-Based Functional Programming

Imperative vs Stream

Streams with Extensions for Reactive Programming

Rx makes Async behavior easy!

(Reactive Pull) Backpressure


What is Rx?• Collections + Time

• A Single Abstraction over data from different sources

• Observer Pattern with Push-based iterators

• Stream Based Functional Programming

• … with Extensions for Reactive Programming

• Async is easy

• Backpressure

Rx Simplifies Complex Work

…Once you understand, of course…


Story Time!


Story Time


Story Time

2012 - MS Open Source’s RX!




• Rx in depth


Key Terms:

An Observable is like Promise ++

(Observable: aka ‘Publisher’)

Observables are most powerful when wrapping external input

An Observable pushes items to Subscribers

Subscribers receive and operate on emitted data

Observables and Subscribers operate on a Scheduler

The following examples use rxJava 1.x

Also, try out rxGroovy


Groovy• Dynamic Language for the JVM

• Less Verbose (Reduce Java Boilerplate)

• Ruby/Python - esque Collections

• Run Time Meta Programming

• Optionally Typed

• AST Transformations

• Powerful Annotations

• Multi - Inheritance via Traits and @DelegatesTo


Basic Usage

Thankfully, there are shortcuts

Streams are Composable


You can get much power from 5 functions• filter

• map

• reduce

• groupBy

• flatMap

First Mental Leap: An Observable of Observables

–Johnny Appleseed


Let’s get a little crazy


Hot vs Cold

Cold Observable: finite data, on demand Hot Observable: infinite data, as it’s ready

Cold Observable: only starts emitting data on .subscribe () Hot Observable: emits data whenever it’s ready


Asynchronous Streams


BackPressure


Can only Mitigate Hot Streams• throttle

• sample

• window

• buffer

• drop


Stream Interaction

Don’t Unsubscribe from Observables

Programmatically complete them when another Observable fires


AutoComplete Requirements• Wait 250 ms between keypresses before querying

• If no keys are pressed, no query

• Successful queries should render movies

• Any new queries should kill in-flight queries

Pretty Great, But what’s going on?

keyPress stream creates a sub- stream which is reacting to subsequent data from the parent


Questions?

Ok, what about 2.0?

API hasn’t changed much


RxJava 2.0• Completely rebuilt for Reactive-Streams.org 1.0 spec

• Decreased resource usage

• No nulls allowed

• New Publisher types

http://Reactive-Streams.org


Publishers• Observable - classic. In 2.0: no Backpressure

• Flowable - new to 2.0. Use this for Backpressure

• Single - only one item will be emitted or signal error

• Completable - only signal success or error

• Maybe - new to 2.0: one item emitted, signal success, or signal error (think, Promise)




• Rx In Depth

• An Overview of JVM options


Intra-Service Options

RxJava, Obviously


ReactiveX JVM Family• rxJava

• rxJavaFX

• rxGroovy

• rxClojure

• rxKotlin

• rxScala

• And many more (beyond JVM): https://github.com/ReactiveX

https://github.com/ReactiveX


Akka & Akka Streams• Library

• Definition of Reactive System

• Created by LightBend

• Actor-Based Concurrency

• Implemented Streams on Top of Actor Model

–Johnny Appleseed



• Pivotal Project

• Library

• Reactive Streams

• Built on LMAX Ring Buffer / Disrupter

• Multiple libraries to extend Disruptor in multiple ways

Java 9 will have Flow


Framework Level


• High Performance Web Framework

• Non-Opinionated

• Non-Blocking Network Stack

• Built on Reactive Streams, Netty, Java 8, Guice

• Deterministic Asynchronous Execution

Take a Look at Ratpack

http://www.infoq.com/presentations/ratpack-2015


Includes rxRatpack module, but we’ll talk about that later

… specifically, Spring 5


Spring Web Reactive vs Spring Web MVC


Spring 5• spring-web-reactive instead of spring-mvc

• Same @Controller programming model

• Different underlying API

• Based on Project Reactor

• Can run within Tomcat, Jetty, or Netty (e.g. can fallback to use servlets)


Play Framework• Part of the Lightbend family

• Built on Akka and Netty

• Async I/O

• Lightweight and stateless

• Encourages RESTful design

• Focus on JSON


Vert.X• Event-driven & Non blocking

• lightweight, non-opinionated, and modular

• integrates with rxJava

• support for additional languages (like JS and Ruby)

• Can be used as a library embedded in an existing app


Demo Time


Any Questions?

Thank You!

@svpember [email protected]


mailto:[email protected]


More Information• Reactive Groovy & Ratpack Demo: https://github.com/spember/reactive-movie-demo • Jafar Husain: RxJS: https://www.youtube.com/watch?v=XRYN2xt11Ek • Reactive Streams Spec: http://www.reactive-streams.org/ • Reactive Manifesto: http://www.reactivemanifesto.org/ • Akka: http://akka.io/ • rxJava / ReactiveX libraries: https://github.com/ReactiveX • Ratpack: http://ratpack.io/ • Reactor: https://github.com/reactor/reactor • The Introduction to Reactive Programming you’ve been missing: https://gist.github.com/staltz/868e7e9bc2a7b8c1f754 • Martin Fowler: Stream / Pipeline programming: http://martinfowler.com/articles/refactoring-pipelines.html • Or Just on Groovy (Groovy the Awesome Parts): http://www.slideshare.net/SpringCentral/groovy-the-awesome-parts • Ratpack Web Presentation: http://www.infoq.com/presentations/ratpack-2015 • Advanced RxJava Blog: http://akarnokd.blogspot.com/ • Martin Fowler LMAX breakdown: http://martinfowler.com/articles/lmax.html • Reactive Web Apps with Spring 5: https://www.youtube.com/watch?v=rdgJ8fOxJhc&feature=youtu.be

https://github.com/spember/reactive-movie-demo

https://www.youtube.com/watch?v=XRYN2xt11Ek

http://www.reactive-streams.org/

http://www.reactivemanifesto.org/

http://akka.io/

https://github.com/ReactiveX

http://ratpack.io/

https://github.com/reactor/reactor

https://gist.github.com/staltz/868e7e9bc2a7b8c1f754

http://martinfowler.com/articles/refactoring-pipelines.html

http://www.slideshare.net/SpringCentral/groovy-the-awesome-parts


http://akarnokd.blogspot.com/

http://martinfowler.com/articles/lmax.html

https://www.youtube.com/watch?v=rdgJ8fOxJhc&feature=youtu.be

Images• Empty Pool: http://www.wtok.com/home/headlines/Water-Problems-205987121.html

• Juggling: https://en.wikipedia.org/wiki/Juggling

• Directing Traffic: https://www.flickr.com/photos/tracilawson/3474012583L

• LMAX Disrupter: http://martinfowler.com/articles/lmax.html

• Mailman: thebrandtstandard.com/2013/02/09/u-s-post-office-to-end-saturday-letter-delivery-this-summer/

• Actors Diagram: https://blog.codecentric.de/en/2015/08/introduction-to-akka-actors/

• Cheetah: www.livescience.com/21944-usain-bolt-vs-cheetah-animal-olympics.html

• Dominoes: https://www.flickr.com/photos/louish/5611657857/sizes/l/in/photostream/

• Spartans: www.300themovie.com/

• Stampeding Buffalo: news.sd.gov/newsitem.aspx?id=15164

• Turtles (Cosmic): http://synchronicity313.deviantart.com/art/Turtles-All-The-Way-Down-68160813

• XkCD turtles: https://xkcd.com/1416/

• Simpsons “Old Coot”: http://simpsons.wikia.com/wiki/Category:Grandparents

• Meditation: http://i.huffpost.com/gen/1405484/images/o-MEDITATION-facebook.jpg

• Death Star Architectures: http://www.slideshare.net/adriancockcroft/monitorama-please-no-more

http://www.wtok.com/home/headlines/Water-Problems-205987121.html

https://en.wikipedia.org/wiki/Juggling

https://www.flickr.com/photos/tracilawson/3474012583

http://martinfowler.com/articles/lmax.html

http://thebrandtstandard.com/2013/02/09/u-s-post-office-to-end-saturday-letter-delivery-this-summer/

https://blog.codecentric.de/en/2015/08/introduction-to-akka-actors/

http://www.livescience.com/21944-usain-bolt-vs-cheetah-animal-olympics.html

https://www.flickr.com/photos/louish/5611657857/sizes/l/in/photostream/

http://www.300themovie.com/

http://news.sd.gov/newsitem.aspx?id=15164

http://synchronicity313.deviantart.com/art/Turtles-All-The-Way-Down-68160813

https://xkcd.com/1416/

http://simpsons.wikia.com/wiki/Category:Grandparents

http://i.huffpost.com/gen/1405484/images/o-MEDITATION-facebook.jpg

http://www.slideshare.net/adriancockcroft/monitorama-please-no-more