EyeOne.com

Real-time data streaming

Some data is only worth something while it is fresh: a price, a score, a position on a map, the state of a machine. Instead of letting every browser ask again and again, the server keeps a connection open and pushes each change the moment it happens. That is how broker sites stream prices, and it is how the multiplayer games on this site work.

WebSocketServer-Sent EventsSocket.IOuWebSockets.jsSignalRSpring WebSocket & STOMPCentrifugoAbly, Pusher & PubNubRedis pub/subKafkaNATS
Get started

The tools

Live prices, dashboards and notifications pushed to the browser with WebSocket, Server-Sent Events, Socket.IO, SignalR, Redis pub/sub, Kafka and NATS.

WebSocket

One open two-way connection per browser. The server streams each update as it happens and the client can answer on the same channel. Our game servers run on it.

Server-Sent Events

A simpler one-way stream from server to browser over plain HTTP, with automatic reconnection. Often all a price feed, a dashboard or a notification list needs.

Socket.IO

A Node.js library on top of WebSocket with rooms, acknowledgements and fallback transports. A quick, well-known way to add live features to a NestJS or Express backend.

uWebSockets.js

A very fast WebSocket server for Node.js, for when one machine has to hold a very large number of connections.

SignalR

Microsoft's real-time framework for .NET. It chooses the best transport by itself and fits naturally with ASP.NET Core and Azure.

Spring WebSocket & STOMP

WebSocket support in Spring Boot, with STOMP for topics and subscriptions, so Java services publish to browsers through the same messaging model they use internally.

Centrifugo

A stand-alone real-time server written in Go that scales to very many connections. Your backend publishes over a simple API and Centrifugo handles the clients.

Ably, Pusher & PubNub

Hosted real-time networks for teams that do not want to run the infrastructure themselves: global delivery, presence and message history as a service.

Redis pub/sub

The simplest message bus: each update is published once and Redis fans it out to every WebSocket server, which passes it on to its own clients.

Kafka

A durable event log for high volumes. Every tick is stored in order, several consumers read at their own pace, and a stream can be replayed after a failure.

NATS

A light, very fast messaging system for services that need low latency and simple operations, with optional persistence.

How we work with them

SSE one way, WebSocket two ways

A one-way feed uses Server-Sent Events. WebSocket is for when the client talks back, as in a game, a chat or an order screen.

A message bus behind the servers

Each update is published once to Redis, Kafka or NATS and fanned out to as many WebSocket servers as the audience needs, so capacity grows by adding servers.

Throttled, small updates

Updates are limited to a few per second per item and sent as small differences, in binary when it pays off, so a busy market does not flood a phone.

Subscribe only to what is on screen

A client subscribes to the items it is showing and unsubscribes when the user moves on, which keeps traffic and cost down.

Planning a project on this stack?

Tell us what you are building and we will tell you how we would approach it.

Start a conversation