Site Reliability Engineering

Site Reliability Engineering is a practical engineering guide to designing, operating, and sustaining production systems that users can rely on. It starts from a single observation: reliability is not an aspiration, it is an engineering discipline, and like every engineering discipline it has measurable targets, trade-offs, and practices that determine whether it works. The book walks through the full practice — the case for SRE and how it differs from traditional operations, service level objectives and error budgets, monitoring and alerting that respects the humans on call, incident response and on-call practice, blameless postmortems and organisational learning, capacity planning and demand forecasting, change management and release engineering, toil reduction and automation, distributed systems reliability, chaos engineering and game days, observability engineering, reliability economics, SRE in the cloud era, and the cultural practices that make the discipline stick. It covers the failure modes that quietly destroy SRE programs: an SLO nobody enforces, an error budget that is exhausted and ignored, an alerting rule that fires a thousand times a day, an on-call rotation that burns out its engineers, a postmortem that identifies human error as the cause and ships no action items, a chaos experiment that takes down production, an observability stack that costs more than the infrastructure it observes. Each is presented with the failure, the countermeasure, and the operational tradeoff. Fourteen chapters, 45,000 words. Real Prometheus rules, SLO definitions, postmortem templates, chaos experiment manifests, and DORA metrics. Written for engineers, SREs, platform teams, and engineering leaders who need reliability to be a discipline rather than a slogan.