The Perverse Incentives of Reliability
summary
Claim: companies only reward reliability after SEV1s, so focus on keeping restorations shorter than the downtime tolerance window rather than chasing impossible “infinite uptime.” Tools that prove useful are automated roll back and try again, traffic quarantine/drain, load shedding, and regular rollback drills; these work when exercised. Surprising metric: building a safety-net approach often requires roughly three times the SRE headcount. Tactics that buy funding are partnering with sales/legal or product/support and using curiosity and incident rituals to pull devs toward reliability.
AI-written summary · about
chapters (9)
- 0:00Perverse reliability incentives
- 3:09Uptime failure impact
- 7:31Bystander effect in dev
- 12:32Extended downtime risk
- 16:27Downtime tolerance windows
- 19:59Mitigation techniques
- 21:47Greed and fear for resources
- 24:33Developers toward reliability
- 26:19Q&A
related talks
29mThe Curious SRE: Cultivating a Mindset for System Resilience
43mThe SRE I Aspire to Be
33mThe Value of Reliability
34mBuilding a Culture of Reliability
45mPanel Discussion: Is Reliability a Luxury Good?
37mSlow and boring: Secrets to building reliable software
30mThe 'Success' in SRE Is Silent
34mReliability and Engineering Requirements