Forum Discussion
Azure SignalR vs Web PubSub for a real-time live score website – which one fits better?
Jamony's comparison is the right one, so I'll just add the bit that matters most for a personal project like this: you don't need an always-on backend at all.
Both services have a serverless mode through Azure Functions, and that's the shape I'd build this in. A timer-triggered function does the polling — one function, every few seconds, calling the score API once — and then pushes only the changes out through the service's output binding. For SignalR that's the SignalR Service output binding for Functions; Web PubSub has the same idea with its own Functions trigger and bindings. Either way there's no app server to keep running and pay for between matches, which is where the cost difference actually shows up for a hobby site: you're paying for function executions and messages, not an idle VM or App Service plan.
On the choice itself, Jamony's steer holds: Web PubSub if you want plain WebSockets and a publish call from any language, SignalR if you want the hub model and its client libraries. And start both on the free tier while you're proving it out — 20 concurrent connections and 20,000 messages a day is plenty to find out whether your polling rhythm and group-per-match design hold up under a real match, before you spend anything on a paid tier.