bakdata / pipeline-atm-fraud

Demo pipeline for ATM fraud detection using streams-bootstrap

Kafka demo pipeline: ATM fraud detection with streams-bootstrap

This is an adaption of the example pipeline for ATM fraud detection using streams-bootstrap. The original by Confluent is written in KSQL and can be found in the ksql-atm-fraud-detection repo. Details can be found in their blogpost

This version differs slightly from the original in that the test accounts are generated using a producer application. In addition, we turned the transactionavroproducer application into a producer application, eliminating the need to consume transactions data from an external producer. These minor changes allow us to have an automated deployment with no external intervention or manual data production.

Usage

Build containers using jib

gradle jib -Djib.to.image=url-to-container-registry.com/streams-explorer-demo-transactionavroproducer -Djib.container.mainClass=com.bakdata.kafka.TransactionAvroProducer
gradle jib -Djib.to.image=url-to-container-registry.com/streams-explorer-demo-accountproducer -Djib.container.mainClass=com.bakdata.kafka.AccountProducer
gradle jib -Djib.to.image=url-to-container-registry.com/streams-explorer-demo-transactionjoiner -Djib.container.mainClass=com.bakdata.kafka.TransactionJoiner
gradle jib -Djib.to.image=url-to-container-registry.com/streams-explorer-demo-frauddetector -Djib.container.mainClass=com.bakdata.kafka.FraudDetector
gradle jib -Djib.to.image=url-to-container-registry.com/streams-explorer-demo-accountlinker -Djib.container.mainClass=com.bakdata.kafka.AccountLinker

Deploy in Kubernetes

helm repo add bakdata-common https://raw.githubusercontent.com/bakdata/streams-bootstrap/master/charts/
helm repo update
helm upgrade --debug --install --force --values values-transactionavroproducer.yaml demo-transactionavroproducer bakdata-common/streams-app
helm upgrade --debug --install --force --values values-accountproducer.yaml demo-accountproducer bakdata-common/streams-app
helm upgrade --debug --install --force --values values-transactionjoiner.yaml demo-transactionjoiner bakdata-common/streams-app
helm upgrade --debug --install --force --values values-frauddetector.yaml demo-frauddetector bakdata-common/streams-app
helm upgrade --debug --install --force --values values-accountlinker.yaml demo-accountlinker bakdata-common/streams-app

Generate test accounts

The account producer application publishes accounts data (stored in a .txt file) into the expected topic.

Generate test transactions

To generate our incoming transactions (legitimate or fraudulent) we are using the transactionavroproducer application inspired by the gess tool. This project's data is derived entirely from the same gess project. The number of incoming transactions is configurable using the variables REAL_TX and ITERATION in our values-transactionavroproducer.yaml file. The first variable specifies how many legitimate incoming transactions must be produced before producing one fraudulent incoming transaction. One iteration contains REAL_TX legitimate transactions and one fraudulent transaction.