SLO-Based Observability For All Kubernetes Cluster Components
Talkat KubeCon + CloudNativeCon North America 2022 · on Friday, 2022-10-28· at 11:55· Portside Ballroom (Room 260)·
summary
Pyrra is an open source SLO tool that deploys as a single controller or binary, creates SLO CRDs and recording/alerting rules for Prometheus (also supports Thanos/Cortex), and groups metrics by handler to produce per-endpoint SLOs. Example targets: API server read/write and kubelet operations with 99%–99.99% windows; observed about 4,000 slow reads out of ~2,000,000 (and roughly 1 error in 4,000,000 operations for a kubelet SLO). Demo triggers a CoreDNS incident by capping CPU and shows multi burn-rate alerts (four severities) and Pyrra creates SLO CRDs and Prometheus rules
AI-written summary · about
chapters (7)
- 0:00SLO-based observability intro
- 4:40Service level objectives fundamentals
- 8:40Pura for SLO management
- 12:47SLOs for Kubernetes components
- 18:45Live SLO monitoring demo
- 21:12Pura and SLO future plans
- 24:26Q&A
related talks
05mPyrra - Making SLOs with Prometheus Manageable, Accessible, and Ea...
40mScaling SLOs with Kubernetes and Cloud Native Observability
25mAn SLO-Driven Approach to Enhance Kubernetes Cluster Reliability
31mServerless Observability: a case study of SLOs
30mFrom Monitoring to Observability: Left Shift your SLOs with Chaos
23mGrafana SLO Demo: Prioritize critical resources with SLO-driven IRM
35mFrom Monitoring to Observability: Left Shift your SLOs with Chaos
22mLook Ma, No Metrics: A Quest Towards Extracting SLOs