Field notes

Observability decisions from real production work

I write about the decisions behind reliable telemetry: what I inspect, what I change and why it matters during an incident.

By Rafael Fahreev, Observability Expert
Incident response7 min read

Your Monitoring Works. Why Does Incident Response Still Start With Guessing?

Dashboards can be green, alerts can be firing, and responders can still have no reliable path from user impact to the failing component. This is how I find the missing links.

22 July 2026Read article →
OpenTelemetry8 min read

How I Take the OpenTelemetry Collector From Proof of Concept to Production

A Collector that exports a demo trace is not yet a production pipeline. I check topology, backpressure, state, routing and the Collector's own telemetry before rollout.

22 July 2026Read article →
Alerting and SLOs7 min read

I Do Not Fix Alert Fatigue by Tuning Thresholds

Threshold tuning can reduce noise for a week. I fix the operating model behind paging: user impact, ownership, severity, action and a deliberate SLO policy.

22 July 2026Read article →