Building an Alerting Strategy

Prerequisites: Metrics and Dashboards with Prometheus; Distributed Tracing.

A rule like error_rate > 1% either fires constantly on harmless blips or misses a real incident buried in a longer window — and no amount of threshold-tuning fixes that, because the underlying problem is the window, not the number. The fix, standardized in Google's SRE Workbook and now the default approach for SLO-driven teams, is burn-rate alerting: instead of asking "is the error rate too high right now," you ask "at this rate, how fast are we consuming our error budget."

This project rebuilds your alerting around that idea, then wires the result into Alertmanager so alerts land with the right severity, on the right channel, without duplicates.