The SRE I Aspire to Be
summary
Argues that SRE work is measurement-first: pick a concrete reliability unit, trade reliability versus cost using error budget math, and move the reliability–cost frontier with engineering (not frantic paging). Gives concrete patterns—RAID (one extra disk drops disk-failure risk from ~1.5% to roughly 1.5%²), partitioning, circuit breaker, retry logic—and metrics (MTBF/MTTR and end‑to‑end success rates). Shows measuring checkout across 25 services can collapse per‑request success to ~97.5%, and that changing the metric and retries raises customer success without extra paging.
AI-written summary · about
chapters (8)
- 0:00Start
- 9:38On ops, user harm, and tradeoffs
- 12:40Engineering reliability
- 16:19The modest reliability engineer toolbox
- 19:05Compound/advanced reliability patterns
- 29:45Measurably optimise reliability vs cost
- 37:15SRE team: a recipe
- 40:35The SRE I aspire to be
related talks
41mSRE in Enterprise
30mImproving Reliability w/ Error Budgets & SRE
30mThe 'Success' in SRE Is Silent
49mWhat SRE Could Be: Systems Reliability Engineering
58mEnterprise Roadmap for SRE
27mGetting Started with SRE
38mImplementing SRE practices: SLI/SLO deep dive
29mThe Curious SRE: Cultivating a Mindset for System Resilience