Redis Streams!

This project is some experimentation with Redis Streams, mainly setting up a producer that xadd's to a stream and a consumer group that...consumes...from the stream.

It's using lettuce, a comprehensive Java Redis client funded by Pivotal and providing three ways of interacting with the Redis API:

Synchronous calls that block
Asynchronous calls that returns Java Futures[1]
Reactive Streams[2] from the Reactor[3] project (also from Pivotal)

Pre-reqs

Grab a very recent Redis 5.x version that supports streams. Install it and get it running locally just using the default settings for now.

This project was built using JDK 11, but should work with JDK 8.

Building & Running

Assuming you've got a recent JDK and javac in your path, this project uses Gradle for dependency and build management. The included wrapper script will download and bootstrap Gradle for you, so all you need is the JDK.

Quickstart

Simply execute the Gradle wrapper with the run task:

$ ./gradlew run

Use gradlew.bat on Windows

You can run the producer ("ticker") or consumer ("taker") individually using gradle as well:

$./gradlew ticker

Building a standalone Jar

I've included a Gradle task for bundling up the project into an "uberjar" to make it easier to use without messing with Java classpaths.

Build the uberjar

$ ./gradlew shadowJar

Run the App

$ ./java -jar ./build/libs/redisfun-0.1.0-SNAPSHOT-all.jar

The exposed entrypoints:

main App: io.sisu.redisfun.App
producer: io.sisu.redisfun.Ticker
consumer: io.sisu.redisfun.Taker

Which you can easily run via:

$ ./java -cp ./build/libs/redisfun-0.1.0-SNAPSHOT-all.jar \
  io.sisu.redisfun.Ticker

Todo's

Explore the following:

stand-alone consumers that use reactor-style consumption models
more complicated POJO storage...can I stick POJOs in the stream?
compare with how this stream-based messaging approach would differ from a "brokerless" model like from zeromq

Footnotes

(These never render well in Github.)

1 2 3

indynz / redisfun