Spinnaker Summit 2020

Canary deployments are a great way to reduce risk while maintaining speed. But the fact is, even the most cutting-edge organizations leveraging this approach still experience frequent production issues. So what’s missing? Historically, engineering teams have assessed canary release quality through the lens of performance and other high level gauges, for example, “Is my service up or down?” or “ Is it slow?” But those metrics don’t tell the full story. To truly have confidence in your release, you need to drill down to the code level to understand what’s new, what’s changed and what’s critical. That’s why we’ve identified 4 key error metrics that you can leverage in your Spinnaker canary analysis to make better pass/fail judgments and ensure the reliability of your applications. In this session, we’ll delve into why traditional judgment metrics have fallen short, then define and walk through each of the 4 error metrics, including real-world stories, key data sources and best practices for configuring thresholds.