Look Ma, No Metrics: A Quest Towards Extracting SLOs
summary
Project shows SLOs can be derived without code changes by converting logs into SLIs: a log-exporter queries the central log store, exposes Prometheus-style metrics, and Grafana dashboards and alerts run on those metrics. The team targets ~99.x% transaction success over a month, adds a simple anomaly detector using mean and standard deviation over 7/14/28-day windows with +/-3σ, and hits production but faces query explosion per bank, exporter stability crashes and occasional >100% sliding-window artefacts that motivate later instrumentation.
AI-written summary · about
chapters (8)
- 0:00Intro
- 0:55Reliability Management
- 5:43Assumptions
- 7:23Logging
- 9:51Problems
- 11:37Demo
- 14:13Summary
- 17:00QA
related talks
31mServerless Observability: a case study of SLOs
40mScaling SLOs with Kubernetes and Cloud Native Observability
43mOur Experience Tracking and Driving SLO Adoption at Goldman Sachs
28mLean Product Development Through SLOs
30mFrom Monitoring to Observability: Left Shift your SLOs with Chaos
23mGoing from 30 to 30 Million SLOs
32m9 Things You Should Do When Starting to Use SLOs
37mReal World SLOs and SLIs: A Deep Dive