Roadmap215 words
Unit 5 — Instrumentation strategy — roadmap
Unit 5 — Implement an instrumentation strategy
Exam weight: 5–10% — the lightest unit · 2 topics · 8 learning objectives
Small, but it is where you find out whether everything else worked.
| Topic | Objectives | Core idea |
|---|---|---|
| 5.1 Configuring monitoring | 4 | Alert on metrics, investigate in logs |
| 5.2 Analyzing metrics | 4 | Work from the user inward |
The four things to carry
- Application Insights for application questions; VM/Container Insights for infrastructure. Instrumenting only infrastructure tells you CPU is fine while users fail.
- Report p95/p99, not averages. Averages hide the tail and improve as traffic grows.
- Low CPU with high latency means waiting, not computing — a dependency, a lock, a pool.
- KQL shape: source →
where→summarize→order by→take. Filter early, especially on time.
Two failure modes worth recognising
- A trace that stops partway = a service not propagating the correlation context.
- A notification step with no
condition: failed()never fires on failure, and the silence reads as success.
Where it connects
The Query Azure Monitor alerts check (3.3) is this unit feeding back into deployment gating, and postRouteTraffic health checks (3.4) are instrumentation used as a release gate.