Diagnosing Application Performance With EBPF, Pyroscope, and Kubernetes

Learn how to diagnose and optimize application performance in Kubernetes environments using advanced observability tools in this 30-minute conference talk. Discover a comprehensive framework for guaranteeing software performance through kernel-level eBPF profiling with Pixie, continuous profiling with Pyroscope, and OpenTelemetry tracing. Explore a real-world case study demonstrating how to identify performance bottlenecks in Helm chart-packaged applications running in customer clusters, including the process of uncovering inefficient loops and implementing fixes before customers encounter issues. Master the correlation of telemetry data from multiple sources to drill down into implementation hotspots and establish proactive performance monitoring. Gain practical insights into deploying observability tooling in your own clusters, stress-testing applications to their limits, and building a systematic approach to finding and fixing performance regressions faster than traditional metrics allow.