Free to read ยท Three levels

Site Reliability Engineering: user promises, evidence, and recovery

Eighteen developed lessons connect service objectives, budgets, telemetry, response, capacity, overload, releases, and recovery to accountable engineering decisions.

18 lessonsEnglish + ArabicContent reviewed
Start learning

In this course

Study reliability from the user journey through event contracts, objectives, and error budgets, then through alerting, response, and learning. A fictional Library service supplies concrete reservation, durable-confirmation, queueing, and dependency examples while separating availability, latency, and data correctness.

Each of three levels contains six lessons with explained examples, diagnostic mistakes, worked exercises, and official references. Python examples run only on local synthetic values. Every YAML record is a non-executable editorial plan or contract; no service, account, real data, credential, or infrastructure is accessed.

What you will learn

  • Write indicators, objectives, and budgets that reflect user experience and measurement boundaries.
  • Organize alerting, on-call, incidents, diagnosis, and learning with explicit responsibility and evidence.
  • Review capacity, retries, releases, recovery, and failure domains while stating assumptions and limitations.

Your learning path

Three levels

Basic: promises and measurements

Define user risk, trustworthy events, objectives, budgets, telemetry, and toil.

  1. Reliability as a user and engineering decision
  2. SLIs and trustworthy event contracts
  3. SLO windows and latency distributions
  4. Error budgets and release decisions
  5. Telemetry for user impact and diagnosis
  6. Toil reduction and safe automation
Explore this level

Intermediate: response and operational evidence

Design actionable alerts, sustainable response, causal investigation, learning reviews, and capacity reasoning.

  1. Actionable burn-rate alerts
  2. On-call readiness and runbook contracts
  3. Incident command and useful communication
  4. Hypothesis-driven troubleshooting
  5. Postmortems and effective follow-up actions
  6. Capacity, demand, and queue behavior
Explore this level

Advanced: controlled failure and service readiness

Review overload, releases, recovery, reliability testing, failure domains, and production readiness.

  1. Overload, deadlines, and retry amplification
  2. Canaries, compatibility, and change risk
  3. Recovery objectives and data integrity
  4. Reliability tests and bounded experiment design
  5. Dependency contracts and failure domains
  6. Production readiness and reliability planning
Explore this level

Who this is for

  • Developers, operations engineers, and learners who want to understand reliability decisions across a service lifecycle.
  • Teams connecting user objectives with engineering investment and sustainable response.

Before you start

  • Basic understanding of requests, services, and programming; HTTP and Linux familiarity helps but administrative access is unnecessary.
  • Examples can be read without installing monitoring platforms, containers, or databases; calculations use standard Python.

Related technical solutions

The SLO is green while users report failed reservationsTrace population, observation-boundary, aggregation, and data-completeness mismatches before changing a reliability target or dismissing user reports.Retries increase errors and backlog does not drainDiagnose amplification, expired work, and constrained downstream capacity before treating more retries or workers as recovery.Restore completes but application state is inconsistentDistinguish artifact restoration from verified service recovery when records, confirmations, indexes, or external effects disagree.