Real-Time Systems with WebSockets in Production

When orders need to flow from customer to kitchen in under a second, REST APIs aren't enough. Here's how I built real-time order tracking for a cloud kitchen platform.

Architecture Overview

Client → Socket.io → Redis Pub/Sub → Kitchen Display
                  ↓
            PostgreSQL (persistence)

The Challenge: State Consistency

The hardest part wasn't WebSockets — it was keeping order state consistent across multiple kitchen stations viewing the same queue.

// Order state machine
const orderStates = {
  PLACED: ['ACCEPTED', 'REJECTED'],
  ACCEPTED: ['PREPARING'],
  PREPARING: ['READY'],
  READY: ['PICKED_UP'],
} as const;

function canTransition(current: OrderState, next: OrderState): boolean {
  return orderStates[current]?.includes(next) ?? false;
}

In real-time systems, the bug isn't usually in the WebSocket layer. It's in the state management.