Site Reliability Engineering¶

By Sameer Alam · Updated 14 September 2026

Monitoring tools tell you what's happening. SRE practices decide what to do about it: how reliable a service needs to be, when to page a human, how to run an incident, how to learn from it, and how to keep the people on call healthy. This section turns those ideas into concrete, copyable practices.

What You'll Learn¶

• How to define reliability in terms users care about, with SLIs, SLOs, and error budgets
• How to alert on SLO burn rate instead of noisy thresholds
• How to run incidents with clear roles and communication
• How to write blameless postmortems that lead to real improvements
• How to design on-call that's sustainable, and reduce the toil that burns teams out
• How to plan capacity and prove it with load tests

The Reliability Loop¶

flowchart LR
  A["Define SLOs"] --> B["Measure SLIs"]
  B --> C["Alert on burn rate"]
  C --> D["Respond to incidents"]
  D --> E["Postmortem"]
  E --> F["Prioritize fixes with<br/>the error budget"]
  F --> A

Read in This Order¶

1. SLIs, SLOs, and Error Budgets — choosing indicators, setting targets, computing budgets, and error budget policies
2. Alerting on SLOs — multi-window, multi-burn-rate alerts with working Prometheus rules
3. Incident Response — severity levels, roles, communication, mitigation first, and runbooks
4. Blameless Postmortems — timelines, contributing factors, action items that get done, and a template
5. Sustainable On-Call — rotations, handoffs, escalation, alert hygiene, and measuring on-call health
6. Toil and Automation — identifying toil, measuring it, and deciding what to automate
7. Capacity Planning and Load Testing — forecasting, headroom, and load tests with k6

SRE and DevOps¶

DevOps describes a culture of shared ownership between building and running software. SRE is one concrete way to practice it, with engineering-driven reliability targets, budgets, and automation. You don't need a team called "SRE" to use these practices — product teams that own their services benefit from every chapter here.

How This Connects¶

Next¶

Start with SLIs, SLOs, and Error Budgets.