Distributed tracing sounds great until you're drowning in spans from 47 services at 2 AM. Most observability creates more noise than signal—proprietary formats, vendor lock-in, and traces that don't help during actual incidents.

This session explores building production-ready observability using Linkerd's native OpenTelemetry integration. We'll examine how Linkerd automatically generates OTLP-compliant traces and metrics without code instrumentation, analyze telemetry data models and sampling strategies for high-throughput environments, and explore architectural patterns for correlating service mesh metrics with application-level telemetry.

You'll learn about OTEL collector configurations, vendor-neutral observability pipelines, and performance considerations when instrumenting service meshes at scale. The talk covers practical implementation strategies for integrating mesh-native observability with existing monitoring infrastructure while maintaining OpenTelemetry standards compliance.