Sefism Platform Engineering and SRE Cheat Sheet

What this cheat sheet is

The platform engineering and SRE cheat sheet is the revision companion to Sefism's Platform engineering and SRE roadmap. It is for DevOps engineers who already ship to Kubernetes and want the next step, and it covers SLOs and error budgets, deep observability, incidents, production Kubernetes, infrastructure as code at scale, GitOps, internal platforms, capacity, chaos and sane on-call.

Its own description sums up the purpose: the numbers, the queries and the templates you reach for when production is on fire. Each concept says what it is and why it matters, and many include a PromQL query, a YAML manifest, a template or the classic mistake.

Two editions, one per path

The sheet follows the roadmap's path choice:

• Site reliability (the default): SLO maths, burn-rate alerts, PromQL, incident roles, a postmortem template, a Kubernetes troubleshooting flow, capacity sums, resilience patterns and on-call rules, plus production readiness reviews. 76 concepts.
• Platform engineering: the reliability basics every platform must meet, observability as a platform service, namespaces as tenancy, Crossplane compositions, Backstage, golden paths, service scorecards and cost. 79 concepts.

Both editions have the same ten chapters.

Chapter by chapter

1. Reliability basics: SLIs, SLOs and error budgets (7 concepts)

The SLI, SLO and SLA, availability nines and allowed downtime (with the arithmetic), the error budget, the error budget policy and composite availability.

2. Observability: metrics, logs and traces

The four golden signals, RED and USE, counters, gauges and histograms, PromQL basics, recording rules, cardinality, the OpenTelemetry Collector, head and tail sampling and correlating logs, metrics and traces, with observability as a platform service on the platform edition. The PromQL and Collector entries carry snippets.

3. Incident response and postmortems (8 concepts)

Burn rate, the multiwindow, multi-burn-rate alert, incident roles, severity levels, mitigate first and diagnose later, a postmortem template, the blameless postmortem and postmortem action items.

4. Kubernetes in production

A troubleshooting flow, common pod failure states, the PodDisruptionBudget, topology spread constraints, quality of service classes, graceful shutdown, cluster upgrades and admission policies, with namespaces as tenancy on the platform edition.

5. Infrastructure as code at scale

Splitting state by blast radius, versioned modules, plan on pull request and apply on merge, policy as code, drift detection, Terraform and OpenTofu, and Crossplane compositions on the platform edition.

6. GitOps and progressive delivery (7 concepts)

GitOps principles, the Argo CD Application, auto-sync, prune and self-heal, ApplicationSets and app of apps, environment promotion, progressive delivery and secrets in GitOps.

7. Internal developer platforms

Platform as a product, golden paths, the software catalogue, software templates, docs like code, cognitive load and team types, and measuring a platform, plus service scorecards (platform) or the production readiness review (SRE).

8. Capacity, performance and cost (8 concepts)

Little's law, utilisation and queueing, headroom and N+1, load testing, autoscaling at each layer, cost allocation, rightsizing, and spot and commitment discounts.

9. Chaos and resilience testing (8 concepts)

The chaos experiment, blast radius, fault injection in Kubernetes, game days, timeouts, retries, backoff and jitter, circuit breakers, bulkheads, and load shedding and graceful degradation.

10. On-call without burnout (7 concepts)

On-call rotations, escalation policies, actionable alerts, runbooks, toil, the on-call handoff and sustainable on-call load.

Each chapter lists up to three library resources, including the Google Site Reliability Engineering Book, Prometheus Documentation, PagerDuty Incident Response, Argo CD Documentation, Backstage Documentation and Chaos Mesh Documentation.

Rules of thumb and glossary

Eight rules are shared by both editions and each path adds one. Among them: write the SLO before the dashboard; page on symptoms users feel and burn rate, never on CPU alone; in an incident, roll back the last change before you debug it; read the Events at the bottom of kubectl describe before anything else; keep metric labels low cardinality, because IDs belong in logs and traces; if a human changes the cluster by hand, GitOps self-heal should undo it; and run chaos experiments small, with a hypothesis and a kill switch. The platform edition adds a rule about treating the platform as a product and talking to its users before building a portal; the SRE edition adds that every page should link to a runbook.

A glossary of 48 short terms sits alongside, and every concept joins it automatically.

How to use it with the roadmap

• After each stage, read its chapter and redo the key sum or query without looking: the nines table, a burn-rate threshold, Little's law, a PromQL error ratio.
• While building on your kind cluster, keep the Kubernetes and GitOps chapters open. The troubleshooting flow is meant to be followed step by step.
• Before an on-call shift or a technical round for an SRE or platform role, reread incident response and on-call. Being able to explain mitigate-first and a blameless postmortem calmly matters as much as any tool.

Roadmap topics link straight to the concepts that cover them in your edition.

How it works on Sefism

Everything on Sefism is free to learn from. The public page shows all ten chapters with their introductions and opens the first two in full. Keeping the sheet uses credits and opens every chapter in the member area; new members get starting credits when they join. Adding the Platform engineering and SRE roadmap to your dashboard (your first roadmap is free) brings this sheet with it.

Related on Sefism

• How to move from DevOps to site reliability engineering
• Platform engineering vs DevOps
• Status Page with Multi-Region Uptime Checks for Small SaaS Teams, a project built on chapters 1 to 3.
• The DevOps and cloud cheat sheet for the ground this sheet assumes.

FAQ

What should an SRE cheat sheet include? SLO maths, observability queries, incident and postmortem templates, a Kubernetes troubleshooting flow, capacity sums, resilience patterns and on-call rules. This sheet has all of them, plus platform engineering.

Is there a version for platform engineers? Yes. The platform edition adds Backstage, golden paths, service scorecards, namespaces as tenancy and Crossplane.

Does it teach Kubernetes from scratch? No. It covers what changes when a cluster runs real traffic. The DevOps and cloud sheet covers the basics.

Is it tied to paid tools? No. The examples use open source tools such as Prometheus, OpenTelemetry, Argo CD, Backstage and Chaos Mesh, and the ideas carry over to commercial ones.