Why should you care?

HTTP polling wastes bandwidth for live dashboards; Socket.IO handles reconnect and broadcast.

See it live — copy this example

Paste into your MeanVerse project (Angular + Express + MongoDB), then run with ng serve / node / mongosh as noted.

// api/src/realtime/socket.ts
import { Server } from 'socket.io';
export function initSocket(httpServer: http.Server) {
  const io = new Server(httpServer, { cors: { origin: process.env.ANGULAR_APP } });

  io.use(async (socket, next) => {
    const token = socket.handshake.auth.token;
    socket.data.user = verifyAccessToken(token);
    next();
  });

  io.on('connection', (socket) => {
    socket.join(`tenant:${socket.data.user.tenantId}`);
    socket.on('subscribe:alerts', () => socket.join('alerts'));
  });

  return io;
}

// emit to tenant
io.to(`tenant:${tenantId}`).emit('transfer:completed', { id, amount });

What happened?

• Auth middleware validates JWT on handshake.
• join puts socket in tenant room.
• emit targets all connected users in that tenant only.

Practice next

1. Attach Socket.IO to same http.Server as Express.
2. Pass JWT in socket.auth from Angular.
3. Create rooms per tenant and feature.
4. Add acknowledgment callback on client receive.
5. Scale with Redis adapter for multi-pod emit.

Remember

Socket.IO = WebSockets + rooms + reconnect. Authenticate handshake like HTTP. Room per tenant isolates broadcasts.

Fraud alert stream

Risk engine flags transaction; teller UI updates instantly.

Outcome: Socket.IO pushes alert without polling /api/alerts.