Who this is for
Engineers who keep production running and want the vocabulary and the mechanics to do it deliberately rather than reactively. If you have ever been asked "what is our SLO" and answered with a CPU graph, or watched a Prometheus instance fall over after an innocent-looking deploy, this is aimed at you.
No prior SRE experience is assumed. You will get more out of the second half if you are comfortable with containers, YAML, and reading a query language, but each part explains its own prerequisites.
How the series is structured
The order is deliberate and cumulative. The first three parts establish the reliability contract — indicators, objectives, and the error budget that makes them enforceable. Parts four to seven cover measurement and instrumentation, including cardinality, which is the design decision that quietly determines your observability bill. Parts eight to ten add logs and traces and then assemble the whole platform. The last two cover operating it.
Every part ends with runnable configuration, and each one maps to a folder in the companion lab repository so you can follow along on a laptop rather than reading about it.
The curriculum
What you will be able to do at the end
- Define SLIs and SLOs for a real service and defend the numbers to product and to leadership.
- Compute an error budget, write the policy that governs it, and build burn-rate alerts that page on real problems and stay quiet otherwise.
- Read latency distributions correctly, and explain why the p99 on your dashboard is probably being aggregated wrong.
- Instrument a service for the four golden signals and write PromQL that returns what you think it returns.
- Budget metric cardinality before it becomes an outage or an invoice.
- Run Loki, Grafana, Tempo and Mimir, wired together so a metric spike is one click from the trace and the logs that explain it.
- Run an incident with a defined structure, and write a postmortem that changes something.
Start here
If you are new to the topic, read the parts in order. If you already have SLOs and want the hands-on platform work, start at Part 6 and treat the earlier parts as reference. If you are here because something is on fire and your observability bill is the fire, go straight to Part 7 on cardinality.