Site Reliability Engineering
About this book
Google's account of how it runs production systems, written by the engineers who do it. The central ideas are the service level objective as an explicit reliability target and the error budget that follows from it, which turns arguments about release pace into an arithmetic question anyone can check. Chapters cover monitoring and alerting philosophy, on call practice, incident response and postmortems, release engineering, and the effort to keep operational toil below a fixed share of an engineer's time. Later sections deal with distributed systems concerns including load balancing, cascading failures and data integrity. Worth reading for the error budget alone, which gives product and engineering a shared way to trade reliability against speed.
Description via Product Digest.
Topics
Related concepts
- Cost and consumption controlsWhy unbounded consumption is a safety problem before it is a finance one, where the money in an agent run actually goes, and the limits that decide what the worst case looks like before somebody else decides it.
- Prompt Versioning and Deployment: Canary Releases and RollbackVersion prompts with their model, schema and retrieval settings, then deploy through evaluation, canary monitoring and a tested rollback process.
- LLM Observability: Traces, Tool Calls and ReplayTrace LLM requests across retrieval, model calls, validation and tools. Learn what to record, how to protect data and where replay stops being reproducible.
- Corrigibility and controlThe property of a system that lets people correct, interrupt or shut it down, why optimisation works against it, and what keeping it costs in a product.
- How to Evaluate AI Agents: Outcomes, Trajectories and ReliabilityEvaluate AI agents using verified task outcomes, permitted action paths, repeated trials and resource budgets, without requiring one exact tool sequence.
- How to Reduce LLM Cost and Latency: Tokens and CachingCalculate LLM request costs and evaluate prompt caching, shorter outputs, model routing, streaming and parallel calls without hiding quality trade offs.
- How AI Agent Loops Work: ReAct, Planning and Stop ConditionsUnderstand the AI agent loop, compare ReAct with plan and execute designs and set explicit limits on steps, tokens, time and tool actions.
- AI Agent Failure Modes: Loops, Duplicate Actions and Plan DriftDiagnose common AI agent failures and control retries, duplicate actions, invalid tool calls, plan drift and ambiguous tool results.
- LLM Production Monitoring: Quality, Cost, Latency and DriftMonitor LLM applications for answer quality, retrieval failures, tool errors and drift, with meaningful samples, denominators and actionable alerts.
- AI incident responseWhat a team does on the day its system produces something it should not have, the decisions worth making months before that day, and when the law requires the incident to be reported.