Grafana: The open and composable observability platform

Adaptive Traces, part of Grafana Cloud's Adaptive Telemetry suite, introduces a refined approach to tracing in distributed systems by utilizing tail sampling to retain only high-value traces, such as those with errors or elevated latency, thus reducing data storage costs and noise while preserving critical insights. Unlike head sampling, which can miss important traces by making sampling decisions early, tail sampling waits until a trace has collected most of its spans before determining whether to keep or discard it, allowing for more precise capture of performance issues. This method helps teams manage costs and complexity without the need for a self-hosted sampling pipeline. Adaptive Traces also addresses the challenge of skewed metrics by generating accurate metrics from raw trace data before any downsampling occurs, ensuring reliability. It integrates with OpenTelemetry and provides users with flexible policy configurations to tailor their tracing needs, supporting anomaly detection and seamless adaptation within Grafana Cloud's infrastructure. This capability is available across all Grafana Cloud tiers, including the free tier, and is part of a broader suite that includes Adaptive Metrics, Logs, and Profiles to enhance observability and performance analysis.