Sublinear Scaling in Practice: The 1k SRE Project
summary
About 40 people in two locations run roughly 400 Google Ads services and report growing from 187 to 431 services in three years without headcount increases, targeting 1,000. Work focuses on heavy automation: declarative automation driven by intent in version control, actuators and a sequencer invariants layer that safely orchestrates quota, processes and load‑balancer changes. Automated SLO monitoring, load tests, rollouts and policy enforcement reduce manual ops; next steps add programmatic readiness validators, automated incident annotation, auto‑redirects, emergency quota and automatic escalation to developers.
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:52About the Team
- 2:31About Project Work
- 4:05About Sublinear Scaling
- 6:17Imperative Automation
- 10:10Declarative Automation
- 12:13Sequencer
- 16:28Automations
- 19:23Incremental progress
- 22:41Automatic continuous production readiness reviews
- 23:49Automated incident handling
- 25:58Summary
related talks
27mHow Google SRE and developers work together
40mHow Google SRE and developers work together
32mHow Google SRE and Developers Work Together
31mScaling SRE at MailChimp: How We Did It
41mAn Introduction to SRE at Google (with DevOps Focus)
29mStory of Seeking Sustainable SRE
14mWeb Apps, through the Lens of SRE (Site Reliability Engineering)
30mThe Path to SRE