## Core Insight Automation evolves through five levels, from no automation to fully autonomous systems. The ideal is not better scripts — it's systems designed to not need external automation at all. Each level up the hierarchy reduces human involvement and increases reliability, but the biggest leap is from level 4 (internally maintained automation) to level 5 (autonomous), which requires treating infrastructure as a software problem. ## The Five Levels 1. **No automation** — manual operations (e.g., manual database failover) 2. **Externally maintained, system-specific** — individual SRE's personal script 3. **Externally maintained, generic** — shared team script supporting multiple systems 4. **Internally maintained, system-specific** — the system ships with its own automation 5. **Autonomous** — system detects problems and self-heals with no human intervention ## Five Values of Automation 1. **Consistency** — machines don't vary in execution like humans do 2. **Platform** — centralizes mistakes (fix once, fixed forever), extensible, exports metrics 3. **Faster repairs** — reduced MTTR for common faults 4. **Faster action** — machines react faster than humans in well-defined scenarios 5. **Time savings** — decouples operator from operation; savings multiply across all users ## Key Nuance Automation that is crucial but infrequently executed (e.g., cluster failover every few months) is particularly fragile due to extended feedback cycles and bit rot. The highest-leverage investment is in the design phase — building systems that don't need automation rather than automating existing manual procedures. ## Source - [[Site Reliability Engineering - Chapter 7 - The Evolution of Automation at Google|SRE Ch 7: The Evolution of Automation at Google]] by Niall Murphy ## Related Concepts - [[Automation Ownership Must Follow System Ownership]] - [[Autonomous Systems vs Automated Systems]] - [[Toil Definition and Six Attributes]]