How To Use Referee & Kayenta to Canary...
summary
Nike prototypes a production canary pipeline that uses standalone Kayenta plus a custom Referee UI to generate configs, run retrospective analyses, and show pass/fail reports. Contributions include SignalFx and New Relic metric sources and a new Kayenta API endpoint that aggregates repeated judge cycles. Orchestration pipelines deploy baseline and canary server groups, compare golden‑signal metrics using size‑based or weighted routing, and auto‑promote or rollback. This setup catches a real memory spike caused by a middleware agent update and prevents a full production outage.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 3:22Why Kayenta
- 7:41Metrics
- 10:02Routing Strategies
- 12:56Recap
- 15:58Open Source Contributions
- 17:58Standalone Kayenta
- 23:22Summary
- 24:13Stickers
- 24:37Service metrics
- 26:09Retrospective analysis
- 29:00Synthetic traffic
related talks
15mYahoo Case Study: Automated Canary Analysis using Spinnaker Kayenta
01h 04mThe Fine Art of Canary Deployments
34mEnabling Invisible Infrastructure Upgrades with Automated Canary Analysis
51mCutting edge Continuous Delivery: Automated Canary Analysis through Spring based Spinnaker
10mTest in Prod With Canary Everything
14mCanary Deployments for Infrastructure Upgrades
35mFrom Static Rules To Reasoning Platforms: Scaling Intelligent Canary Delivery in 2026
35mCutting-edge Continuous Delivery: Automated Canary Analysis on CF using Spinnaker