SRE, what is it all about?!
summary
A conversational AI platform for TV and mobile gets SRE discipline added: SLIs measured at the load balancer, SLOs set as moving 99% SLO (28-day window) targets, and an error budget used to permit releases or trigger reliability work. Prometheus-style metrics feed an open source dashboard and a generator produces multi-window burn rules. A voice API had a 99.9% latency SLO <1s and burned its budget, prompting blameless postmortems and focused fixes that let the product ship reliably.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:10Failed product launch story
- 3:16Reliability meaning and importance
- 5:01SRE as reliability solution
- 7:09Measuring reliability with SLIs/SLOs
- 10:42Error budget concept and use
- 13:08SLOs versus SLAs impact
- 15:40Adjusting SLOs and error budgets
- 17:05Postmortem culture
- 18:57Outcome of SRE adoption
- 19:17Q&A
