Building Real-Time Kitchen Displays with WebSocket

When a waiter takes an order on a tablet and the kitchen display updates instantly — that's the magic of real-time communication. Here's how we built it for our Café & Restaurant Management system.

The Problem

Traditional restaurant systems rely on printed tickets or periodic polling. This creates delays, lost orders, and frustrated kitchen staff. We needed:

• Instant order delivery from POS to kitchen stations
• Live status updates visible to waiters and managers
• Station-based routing so each prep area sees only relevant items
• Zero message loss even during network hiccups

Our WebSocket Architecture

We chose WebSocket over Server-Sent Events (SSE) because we needed bidirectional communication — the kitchen needs to send status updates back to the POS.

Server Side (FastAPI + WebSocket)

Our FastAPI backend manages WebSocket connections per tenant:

• Each connected client joins a tenant-specific channel
• Order events are broadcast to all clients in the same tenant
• A Redis pub/sub layer ensures messages reach all server instances in a multi-process deployment

Client Side (React)

The kitchen display is a React application that:

• Maintains a persistent WebSocket connection with automatic reconnection
• Groups incoming orders by station (grill, drinks, desserts)
• Displays preparation timers with color-coded urgency levels
• Plays audio alerts for new orders

Handling Edge Cases

Real-time systems need to handle failure gracefully:

1. 1Reconnection — Exponential backoff with jitter when the connection drops
2. 2State reconciliation — On reconnect, fetch the latest state from the REST API
3. 3Offline queue — If the kitchen marks an item as ready while offline, the status update is queued and sent on reconnection
4. 4Heartbeat — Ping/pong every 30 seconds to detect stale connections

Results

After deploying the real-time kitchen display:

• Order-to-screen time: < 200ms (previously 30–60 seconds with polling)
• Lost orders: Zero (previously 2–3 per shift)
• Kitchen efficiency: 25% improvement in average preparation time

Lessons Learned

• Always implement reconnection logic — connections WILL drop
• Use a message broker (Redis) for horizontal scaling
• Test with realistic network conditions (throttled connections, intermittent drops)
• Audio alerts are surprisingly impactful for kitchen adoption

*Want to see our kitchen display system in action? Request a demo and we'll walk you through it.*