Skip to main content

Microservices Architecture

Overview of the architecture in Bellamy Book: a single Backend API (ASP.NET Core) with controller-based modules, plus background workers that consume Kafka events. For a step-by-step walkthrough of the overall architecture, see Understanding the Architecture.

Service Decomposition​

Backend API (monolith)​

Single .NET API with controller-based routing. Main areas:

  • Authentication (api/Authentication) – login, register, refresh, me, user profile, 2FA, impersonation
  • Post (api/Post) – feed, create, get, update, delete, reactions, statistics
  • Comment (api/Comment) – create, reply, get, delete, comment reactions
  • Story (api/Story) – create, feed, view, react, archive
  • Friendship / Follow / friends (api/Friendship, api/Follow, api/friends) – friends, requests, suggestions, block
  • Chat / Call / StrangerChat (api/Chat, api/Call, api/StrangerChat) – messages, conversations, groups, calls
  • Notification (api/Notification) – paged, unread count, mark read
  • Hashtags (api/Hashtags) – search, details, posts, trending, recommended
  • Search (api/Search) – full-text search (Elasticsearch)
  • Blog (api/Blog, api/blog-comments) – blogs, blog comments
  • Admin (api/admin/dashboard, api/admin/backup, api/admin/elasticsearch, etc.) – dashboard, backup, Elasticsearch, analytics
  • License, AiAgent, ApplicationSettings, Ticket, Media, SEO – license, AI agent, settings, tickets, media, sitemap

Background Workers (Kafka consumers)​

Workers run as separate processes and consume Kafka topics:

  • GraphWorker – social graph (Neo4j), friend suggestions
  • ScoringWorker – content scores, recommendation
  • TrendingWorker – trending decay
  • InteractionWorker – engagement processing
  • HashtagWorker – hashtag extraction/indexing
  • ElasticsearchSyncWorker – sync content to Elasticsearch
  • BlogAutoGenerationWorker – AI blog generation
  • MediaProcessingWorker – media processing
  • WebSocketWorker – real-time notifications (SignalR)
  • WebPushNotificationWorker – push notifications
  • ChatWorker – real-time chat (SignalR hub), online users, offline messages

Frontend & Admin​

  • Frontend (React + Vite) – social app (feed, profile, messaging, blogs, hashtags, etc.)
  • Admin (React + CoreUI) – dashboard, users, content/blogs, notifications, backup, settings, license, AI agent

Service Communication​

Synchronous​

  • REST between frontend/admin and Backend API
  • SignalR for real-time (notifications, chat)

Asynchronous​

  • API publishes domain events to Kafka
  • Workers consume Kafka topics and update PostgreSQL, MongoDB, Redis, Neo4j, Elasticsearch

Data Management​

  • PostgreSQL – users, posts, comments, reactions, friendships, follows, hashtags, blogs, tickets, etc.
  • MongoDB – stories, chat messages, notifications (and related documents)
  • Redis – cache, sessions, rate limiting, real-time state
  • Neo4j – social graph, recommendations
  • Elasticsearch/OpenSearch – full-text search (posts, blogs)

Deployment​

  • API – containerized (Docker)
  • Workers – one deployment per worker type, scalable
  • Frontend / Admin – static build served by Nginx or CDN

Next Steps​