Skip to Content
Tuan Tran
Observability & SRE

Observability and SRE Exercises

Calculate a request-based error budget

Start with a 99.9% SLO, one million eligible requests, and an observed error rate of 0.05%. Change the SLO or error rate to see what happens when the budget is exceeded. The model uses a single request population and one aggregate error rate. Production alerts calculate separate paired-window burn rates.

Allowed bad requests
1,000
Observed bad requests
500
Remaining budget
500
Budget consumed
50.0%
Burn rate
0.50×

Check your observability and SRE understanding

Which signal measures latency near the slow tail of a request population?
The SLO is 99.9% and the current bad request rate is 1%. What is the burn rate?
Can Prometheus scrape an application endpoint directly?
Does successful canary analysis prove that the service met its 30-day SLO?

Continue with the Payment observability lab.

On this page