There's a Lot of Bad Telemetry Out There - Identifying and Fixing Poor OpenTelemetry Practices

Learn to identify and fix problematic telemetry patterns that create noise, increase costs, and obscure meaningful insights in this 27-minute conference talk from CNCF. Explore real-world examples of bad telemetry including high-cardinality data floods from auto-instrumentation and excessively verbose traces filled with near-useless internal spans that strain collection pipelines. Discover concrete solutions for addressing these issues both at the source and through OpenTelemetry Collector configurations, utilizing filter, attributes, span, and transform processors to improve data quality at the pipeline level. Master OpenTelemetry best practices for creating, filtering, and reshaping telemetry data to cultivate higher-quality, cost-effective, and actionable insights that enhance observability rather than hindering it.