Reference documentation covering cloud architecture, infrastructure as code, containers, delivery pipelines, observability, service level objectives, incident response, and the management layer above them.
Working references rather than tutorials — organised by topic, meant to be returned to. Each document closes with a section on the limitations and failure modes of the thing it describes, and links its primary sources.
| Directory | Contents |
|---|---|
htmls/ |
Standalone single-page HTML. Open htmls/index.html to start — it indexes the set and suggests a reading order. |
mds/ |
The same documents as Markdown, which GitHub renders inline. Start at mds/README.md. |
Every HTML page is self-contained: no server, no build step, no dependency on any file but itself. The only external request is to Google Fonts, and the pages fall back cleanly to system faces without it.
Start with What Good Looks Like — a primer running the full path in one pass, from a sentence describing what a user is trying to do, through a metric and an observability platform, to the page that wakes someone at three in the morning.
- What Good Looks Like — SLIs, SLOs, error budgets, burn rate, alerting and on-call, end to end
- AWS Architecture Reference — core services and the trade-offs between them
- Terraform Operations — state, modules, drift, day-2 operations
- Containers & Orchestration — images, runtime, supply chain, Compose, Kubernetes and ECS
- CI/CD Pipelines — pipeline structure, caching, secrets, deployment strategies, migrations
- Observability Stack — OpenTelemetry, Prometheus, Grafana
- SLOs and Error Budgets — measurement, burn-rate maths, and an operational budget policy
- Incident Response & Failure Modes — triage, failure signatures, command, postmortems
- Reliability & Support Engineering Framework — indicators through to KPIs, OKRs and review cadence
Specific numbers — pricing, service limits, instance generations, tool behaviour — were accurate when written and move independently of these documents. Treat them as the shape of a trade-off rather than as current values, and check the linked primary sources before relying on one.