NOTE: Older Docker images are not available for Apple silicon. If you face this issue, try this WIP branch https://github.com/afedulov/fraud-detection-demo/tree/with-1.15

Fraud Detection Demo with Apache Flink

This demo is related to a three-part blog post series on Advanced Flink Application Patterns:

• Case Study of a Fraud Detection System
• Dynamic Updates of Application Logic
• Custom Window Processing

Requirements:

Demo is bundled in a self-contained package. In order to build it from sources you will need:

• git
• docker
• docker-compose

Recommended resources allocated to Docker:

• 4 CPUs
• 8GB RAM

You can checkout the repository and run the demo locally.

How to run:

In order to run the demo locally, execute the following commands which build the project from sources and start all required services, including the Apache Flink and Apache Kafka clusters.

git clone https://github.com/afedulov/fraud-detection-demo
cd fraud-detection-demo
docker build -t demo-fraud-webapp:latest -f webapp/webapp.Dockerfile webapp/
docker build -t flink-job-fraud-demo:latest -f flink-job/Dockerfile flink-job/
docker-compose -f docker-compose-local-job.yaml up

Note: Dependencies are stored in a cached Docker layer. If you later only modify the source code, not the dependencies, you can expect significantly shorter packaging times for the subsequent builds.

When all components are up and running, go to localhost:5656 in your browser.

Note: you might need to change exposed ports in docker-compose-local-job.yaml in case of collisions.