In this episode of KubeFM, Stephan Schwarz, a DevOps and infrastructure engineer, discusses performance testing Kubernetes workloads. He emphasizes defining performance in context, incrementally testing pod behavior under load, and understanding system-wide bottlenecks. Stephan advises disabling autoscaling, making one change at a time, and documenting each step to understand application limits. He also covers client-side limitations in load testing, the impact of Kubernetes components like ingress controllers, and the importance of monitoring during performance testing. Stephan highlights the role of tracing tools like OpenTelemetry in isolating slow services and the need to account for differences between testing and production environments.
