Design a Chat System

Design WhatsApp or Slack. The hard problems are message ordering, presence detection, offline delivery, and keeping millions of persistent WebSocket connections alive.

What you will learn

• Explain why WebSockets are necessary for chat and when to fall back to long polling
• Design a message ordering scheme that is correct under concurrent sends
• Handle offline users with a durable inbox and push notification fallback
• Implement presence detection without hammering the database

Chat systems are a favourite interview topic because they require designing for state that lives on the wire, not just in the database. A chat connection is a long-lived session. Messages must arrive in order. Users go offline and need their backlog delivered when they return. And you need to know — at low latency — whether someone is online before deciding whether to ring their phone.

This walkthrough covers a WhatsApp-style 1-1 and group chat system.

Functional requirements:

• 1-1 messaging and group chats (up to 500 members)
• Messages are delivered in order and reliably
• Users see the other party's online/offline status (presence)
• Push notification when the recipient is offline
• Read receipts (delivered, read)
• File/image attachments

Non-functional requirements:

• Message delivery latency < 100ms (online recipient)
• Offline delivery: messages must survive indefinitely until the recipient comes online
• High availability — chat downtime during an outage is unacceptable
• Ordering: messages in a conversation must appear in the same order on all clients

Create a free account to read the full chapter — advanced chapters, progress tracking, and quizzes are free with sign-up.