Queue Orchestration POC

This repository demonstrates queue balancing and orchestration using RoadRunner queue API and Temporal server as orchestration platform. Demo is not intended for production usage.

Server Requirements

Make sure that your server is configured with following PHP version and extensions:

• PHP 8.1+, 64bit
• mb-string extension
• PDO Extension with desired database drivers
• MySQL database (or any alternative)
• Temporal Server

Pre-Installations

• Make sure you have a running instance of MySQL, Postgres or etc (this demo can not work with SQLite)
• Install Temporal Server (see an official docker - https://github.com/temporalio/docker-compose)

Installation

Clone the repository.

$ git clone [email protected]:wolfy-j/queue-orchestration.git

Install dependencies:

$ composer install
$ vendor/bin/rr get

Configure your env:

$ cp .env.sample .env

Make sure to update .env with your database credentials!

Configure application and database.

$ php app.php migrate
$ php app.php configure

Application server will be downloaded automatically (php-curl and php-zip required).

Once the application is installed you can ensure that it was configured properly by executing:

$ php ./app.php configure

Additionally, you can tweak configuration options in .rr.yaml.

Starting Application

To start application server execute:

$ ./rr serve

On Windows:

$ rr.exe serve

To view realtime statistic about workers and queue:

$ ./rr workers -i

We recommend keeping this in separate tab for observability.

Running demo

First you have to start the temporal orchestration workflow:

$ php app.php start

To push data into named queue group:

$ php app.php push {group} {payload} -s {count} 

For example (push 1000 messages to bob group):

$ php app.php push bob "hello world" -s 1000 

Make sure to push to different groups to observe the behavior.

To flush counts in case of any error:

$ php app.php flush

How it Works

Demo is fully build Spiral framework, it contains two main parts.

Queue API

RoadRunner application server exposes low-level API to manipulate with Queue brokers, such as create, destroy, consume and pause. This application demo uses in-memory queue provider, but it's capable running on RabbitMQ, Amazon SQS, Beanstalk, Kafka, NATS and etc.

You can find API calls in here.

Workflow

In order to manage the queue application implements RouteWorkflow.

This workflow performs a simple logic of checking the optimal routing configuration every second and then reconfiguring system according to given configuration.

Current logic pushes all message groups with count > 100 to dedicated queue with lower priority (or prefetch), the rest processes in the default queue.

Feel free to implement your own balancing logic.

License:

MIT License (MIT). Please see LICENSE for more information. Maintained by Wolfy-J.