Scaling SRE at MailChimp: How We Did It
summary
Mailchimp describes how an SRE team scales from 4 to 17 engineers while keeping focus on concrete practices: adopt four golden signals for latency, traffic, errors and saturation; implement telemetry with logs (ELK, Sumo Logic, Splunk, CloudWatch), metrics via Prometheus and Grafana, profiling and distributed tracing with OpenTelemetry, Zipkin/Jaeger and X‑Ray; start with a flagship service, build SLIs/SLOs/SLAs, run quarterly readiness reviews, avoid owning every ops task and prioritize where SRE work reduces toil and MTTR.
AI-written summary · about
chapters (19)
- 0:00Start
- 0:34Background about Myself
- 2:19What Is sre Cert Reliability Engineering
- 3:14Working through Incidents
- 6:31What Happens with Your First Sre
- 6:45How Do We Go from Zero to Hero from an Sre Perspective
- 7:16Read the Sre Books from Google
- 7:55Building Secure and Reliable Systems
- 10:23Logging
- 11:32Metrics
- 12:43Profiling
- 13:47Distributed Tracing
- 15:19Capturing the Four Golden Signals
- 18:38Saturation
- 20:33Response Time Histogram
- 21:41Service Level Agreements
- 23:07Prioritizing
- 25:58Learning When To Engage and Disengage with Your Product Teams
- 27:52Growth
related talks
27mSublinear Scaling in Practice: The 1k SRE Project
30mThe Path to SRE
25mUnleashing the potential of SRE for modern operations
24mHow to Destroy an SRE
25mSRE Guiding Principles & Transformational Framework
25mSRE in the Third Age
27mMean Time To SRE: A DevOps story from adidas
29mStory of Seeking Sustainable SRE