From Counter.Inc() to a billion series

Add a counter. Two lines and you can finally see how often your service fails.

Then you add a label to tell failure reasons apart and the counter becomes genuinely useful. You can now see why it's failing.
So you start adding more labels, one for the kind of user, one for the value of the operation. Each one seems reasonable and enables rich queries and alerts.

This talk follows that counter all the way down: from instrumenting a Go service, to how Prometheus scrapes and stores your data, to building a horizontally scalable distributed metrics database as you go from just one counter to billions of them.

You don't need to know anything about observability to attend this session!