mockstagram

Instagram clone web service project built on a microservices architecture.

Target Architecture

How to run

• Run ./start.sh script to run all services with docker compose.
  • It will ask permission once to add domains(mockstagram.com and image.mockstagramcdn.com) on /etc/hosts file.
• Open Browser and type mockstagram.com on address bar. Then, the site will open.

Directory Structure

The directories are organized by domain, and each microservice’s directory may include subdirectories named api, batch, consumer, and domain (with exceptions for admin, web, and media-server).

• /admin
  • Back-office service (backend & frontend)
• /chat
  • Chat service
• /comment
  • Service responsible for managing comments
• /content
  • Service handling content uploads and retrieval
• /media-server
  • Service for image and video uploads, encoding, and related tasks
• /notification
  • Service for in-app notifications
• /push
  • Service for push notifications and scheduling
• /reaction
  • Service managing user reactions (e.g., likes, saves)
• /recommend
  • Service providing APIs for recommendation feeds
• /search
  • Service responsible for search functionality
• /user
  • Service managing accounts and user-related features
• /web
  • React-based web client and web BFF

Assumptions

• Given that Instagram is a global service used by users worldwide, each type of database has a primary DB in one region, with replica DBs in several other regions. Servers handling user requests are assumed to be geographically distant from the primary DB. To simulate this, I plan to introduce intentional network latency and design and implement the system with these considerations in mind.

QnA

• Why aren't all the microservices based on Spring Boot and Gradle managed within the same gradle project?
  • According to Conway’s Law, the structure of a system tends to mirror the structure of the organization and the services used here will be managed by multiple teams in the real world. If a mono-repo is not being used, each service will be located in a separate remote repository managed by its respective team. Some teams may manage multiple services within a single repository within a same gradle project. This can vary depending on the organizational structure and ownership, so I'd rather assume that each service is stored in a separate remote repository.
• Why is BFF (Backend For Frontend) developed using Typescript and Node.js?
  • The concept of BFF (Backend For Frontend) originally emerged to be managed by the team developing and operating a specific client application, which is why it uses the same technology stack as the web client.
• Why do all the microservices share a single database even though it's a microservices architecture?
  • For the sake of simplicity and convenience in development and testing, the databases used by each microservice were configured as separate logical DBs within a single physical DB. However, each service was assigned a unique account, granting access permissions exclusively to its own logical DB.