Introduction to Adaptive Telemetry Processor (ATP)

The Adaptive Telemetry Processor (ATP) is an intelligent metric filtering processor for the New Relic Distribution for OpenTelemetry (NRDOT) Collector. ATP dynamically monitors system and process metrics, filtering out low-value telemetry based on configurable thresholds while preserving critical visibility during performance events.

Why ATP matters

Traditional monitoring approaches create several challenges:

• Data overload: High-volume telemetry data from all processes increases costs without proportional value
• Visibility gaps: Stateless filtering creates "data gaps" where processes constantly appear and disappear from telemetry as they fluctuate around thresholds
• Lack of context: Important processes may be missed during brief but critical resource spikes
• Manual tuning: Static thresholds require constant adjustment as system conditions change

What is ATP

Instead of simple threshold-based filtering, ATP combines:

• Process-based sampling: Monitor specific processes as per your requirement.
• Multi-metric composite scoring: Combine multiple metrics with configurable weights for holistic assessment
• Dynamic threshold adjustment: Automatically adapt thresholds based on historical baselines
• Anomaly detection: Detect sudden metric changes and preserve anomalous data
• Persistent state management: Maintain context across collector restarts
• Low overhead: <2% CPU impact, 10-50MB RAM usage

How it works

ATP processes telemetry data through an intelligent filtering system:

• Process identification: Monitors all processes for resource usage patterns
• Threshold evaluation: Compares current usage against configured or dynamic thresholds
• Retention tracking: Maintains state for processes that exceed thresholds
• Intelligent sampling: Applies advanced logic including anomaly detection and multi-metric evaluation
• Data forwarding: Sends only relevant telemetry data to New Relic

Related resources