Inband Telemetry Flow Aggregation for Collector Load Reduction

The patent combines multiple individual packet telemetry reports into aggregated flow-based telemetry metrics. Instead of reporting each packet's telemetry data separately to the collector, the system aggregates telemetry data from multiple packets belonging to the same flow and reports combined metrics (such as average latency, packet loss rates, or other statistical measures) periodically. This merging approach maintains measurement precision by preserving flow-level telemetry information while dramatically reducing the collector's processing burden by consolidating numerous individual reports into fewer aggregated reports.

The patent applies preliminary aggregation action at the telemetry reporting node before data is transmitted to the collector. The system pre-processes telemetry data by grouping packets into flows and computing aggregated metrics (such as average latency, jitter, packet loss percentages, or other statistical measures) locally at the reporting node. This preliminary action reduces the volume of data that needs to be transmitted to and processed by the collector, while still maintaining complete and accurate flow-level telemetry information. The aggregation is performed in advance, so the collector receives ready-to-analyze summarized data rather than raw individual packet telemetry records.

The patent changes the parameter representation of telemetry data from individual packet-level measurements to flow-level aggregated statistical parameters. Instead of transmitting raw telemetry data for each packet (which would preserve complete detail but overwhelm the collector), the system transforms the data into meaningful statistical parameters such as average latency, maximum latency, minimum latency, jitter (standard deviation of latency), packet loss rates, and other flow-level metrics. This parameter transformation maintains the essential telemetry information needed for network performance monitoring and analysis while dramatically reducing data volume. The aggregated parameters capture the critical characteristics of packet flows without requiring transmission of every individual packet's telemetry data.