Event Driven Architecture (EDA) - Minimal PoC

This project serves as a PoC in implementing EDA using Redis as Event Manager (Broker)

PoV

• Assume we have 3 microservices in our architecture and we want them to talk to each other.
• There are so many ways to do that but lets explore the easiest approach which is realtime and fast.
• Redis can make this happen with easy by streaming events.

producer   ==> service A
consumer-1 ==> service B
consumer-2 ==> service C

Components

• Docker/Compose - package the service
• Redis - act as events manager (broker)
• Microservices (eg. Python) - actual servies (any language)

- Redis offers alot of features and functionalities but in this project we'll use it to tackle:
1. Real-time data store
- With this PoC it covers the scenario where real-time data are exchanged within microservices.
- Publisher and Subscriber flow to generate and consume data in real-time

2. Streaming & messaging
- The stream data type enables high-rate data ingestion, messaging, event sourcing, and notifications.
- Ensure to miss no data from streaming services / APIs.

Project structure

With redis streaming stack to exchage data on real time - Event Driven Architecture (EDA)

• Its suitable for multiple services (more than 2 microservices)
• Can be complex to implement as a number of microservices increases
• Easy to scale services horizontally

.
├── consumer-1
│   ├── consumer-service-1.py
│   └── requirements.txt
├── consumer-2
│   ├── consumer-service-2.py
│   └── requirements.txt
├── Makefile
├── producer
│   ├── docker-compose.yml
│   ├── Dockerfile
│   ├── publisher-service.py
│   └── requirements.txt
└── README.md

4 directories, 10 files

Run

• Run the producer service (service A) by running make run in the folder producer. It will run on http://localhost:5000/
• Run consumers consumer-2 and consumer-2 by navigate into respective folders and run python3 consumer-service-1.py
• Make sure you've installed all requirements.txt
• Consumers in this PoC will stream data by printing them, so leave them running (listen for an event from producer)
• Hint: you can build them separately as well

Todo On Prod

• Think of security and prod configs
• Service authentication
• Explore other redis configs eg. data caching
• Prod level deployment

Thanks

• To me (maen08)