Chat App

A small system-design challenge to practice and tinker with { "docker", "GoLang", "Ruby-on-Rails", "MySQL", "RabbitMQ", "ElasticSearch" }

What are we serving today? 🍽️

• Business Requirements
• The Approach
• Run it
• Remaining Enhancements
• 🚧 Work in progress 🚧

Business Requirements

Our goal is to build a scalable & robust chat app that contains the following requirements:

• Business Entities:
  1. Application (aka. User), can have many chats.
  2. Chat, can have many messages.
  3. Message.
• Chats and Messages have a number that increments for each Application and Chat respectively.
• Client can Create/Read/Update Applications/Chats/Messages, and search messages.
• Client can search Messages.

The Approach

The system is designed as a microservices architecture, with the following components:

1. Main App (Ruby on Rails): Accepts HTTP requests for the the app resources. Serves the Application resource requests, and redirects other requests back to Go service.
2. Chats/Messages Service (Go): Handles the Chat/Message creation/update by generating a number for the resource and publishing a RabbitMQ-message. And reading Chat/Message goes directly to MySQL.
3. RabbitMQ: A message broker to act as a medium between the Rails/Go services and MySQL for asynchronous communication.
4. Ruby-Whenever (Cron): A background processes to count Chats and Messages periodically and persist results in MySQL as a method of caching.
5. Consumer (Ruby): Consumes Chats/Messages published RabbitMQ-messages for resources creation/update and serve send a bulk insert/update in MySQL.
6. Elasticsearch: Searching Messages.
7. Redis: For caching chats and messages numbers.
8. RDBMS (MySQL): Stores Applications/Chats/Messages.

System Architecture Diagram

Run it

To run the system locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/ShadyZekry/chat-app.git
   cd <repository-directory>

2. Build and Run Docker Containers:

   docker-compose up --build

3. Access the system (for inspection):

   1. Create Application

      curl --location 'localhost:3000/api/v1/applications' \
           --form 'name="application-name"'

   2. Get Application

      curl --location 'localhost:3000/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f'

   3. Update Application

      curl --location --request PUT 'localhost:3000/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f'

   4. Create Chat

      curl --location 'localhost:8080/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f/chats' \
      --form 'name="name2"'

   5. Get Chat

      curl --location 'localhost:8080/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f/chats/4'

   6. Update Chat

      curl --location --request PUT 'localhost:8080/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f/chats/4' \
      --form 'name="test"'

   7. Create Message

      curl --location 'localhost:8080/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f/chats/4/messages' \
      --form 'name="name2"'

   8. Get Message

      curl --location 'localhost:8080/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f/chats/4/messages/3'

   9. Update Message

      curl --location --request PUT 'localhost:8080/api/v1/applications/1d763d85-1b08-4941-97fe-e6937931054f/chats/4/messages/3' \
      --form 'name="test"'

   10. Check the internals (For inspection not as APIs) - Main App: http://localhost:8080 - Redis: http://localhost:6379 - RabbitMQ admin-panel: http://localhost:15672

       • Username: guest
       • Password: guest - Elasticsearch: http://localhost:9200 - MySQL: localhost:3306
       • Username: root
       • Password: password

Remaining Enhancements

API Security

• Rate Limiting: Implement rate limiting to prevent abuse and ensure fair usage.

Testabillity

• Adding Specs: Add unit tests to the codebase to ensure reliability and maintainability.

Resiliency

• Retry Mechanism: Add retry mechanisms for transient failures in asynchronous communication between microservices.

Other Enhancements

• Monitoring and Logging: Enhance monitoring and logging to gain better insights into the system's behavior and performance.
• Data Coalescing: Combine multiple Create/Update requests on the Consumer lever to reduce the hit on MySQL.

🚧 Work in progress 🚧

• Elastic search is not yet fully configured.
• Application goes directly to MySQL. The idea behind that decision is that Application creation should not have that huge impact on the system relative to Chats/Messages.