Flow Traffic Estimation Using Interface Stats and Packet Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods cannot accurately estimate the grand total amount of traffic for each flow without collecting header samples of all packets, leading to increased network load when all packets are sampled, or inability to calculate flow-level traffic from interface-level statistics.
Innovation Solution
A traffic monitoring device that acquires interface and flow statistical data from an external storage device, using these data to calculate the grand total amount of traffic for each flow based on sampled header samples and telemetry data, without requiring full packet sampling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If header samples of all packets are collected to calculate the grand total amount of traffic for each flow, then measurement precision is improved, but network load increases significantly
Solution Approach 1:
The patent segments the traffic measurement approach by separating interface-level statistics (collected via telemetry) from flow-level statistics (calculated from sampled packets). This allows the system to use lightweight interface stats for overall traffic estimation while using selective packet sampling for detailed flow analysis, thereby reducing network load while maintaining measurement accuracy.
Solution Approach 2:
The patent introduces interface statistical information as an intermediary element that bridges the gap between network traffic and flow-level details. By using interface stats from telemetry as a proxy for overall traffic patterns, the system can estimate grand total traffic without requiring complete packet sampling, thus reducing network load while preserving measurement precision.
2Object-generated harmful factors
If interface statistical information is collected from network devices to obtain grand total traffic, then network load is reduced, but the ability to calculate flow-level traffic is lost
Solution Approach 1:
The patent merges interface-level statistical information (from telemetry) with flow-level statistical information (from sampled packets) to achieve both low network load and preserved flow-level detail. The system combines these two types of data to calculate grand total traffic for each flow, thereby eliminating the trade-off between network load reduction and information preservation.
Solution Approach 2:
The patent makes the statistical information collection system multi-functional by enabling it to serve both interface-level monitoring (for overall traffic overview) and flow-level analysis (for detailed packet inspection). This universal approach allows the same infrastructure to provide both high-level summaries and detailed breakdowns without requiring separate systems.
3Object-generated harmful factors
If sampled header samples are used to calculate flow statistical information, then network load is reduced, but the ability to accurately estimate grand total traffic is lost
Solution Approach 1:
The patent changes the parameter used for traffic measurement from individual packet header samples to interface statistical aggregates. By using interface stats (which represent aggregated traffic at the interface level) in combination with sampled flow information, the system can accurately estimate grand total traffic while maintaining reduced network load.
Data Source
AI summary
A traffic monitoring device (20) acquires IF statistical data that is statistical information for each IF acquired from a network device and flow statistical data that is statistical information calculated from a sampled packet, from an external storage device that accumulates the IF statistical data and the flow statistical data, and calculates a grand total amount of traffic for each individual flow on the basis of a grand total amount of traffic for each IF calculated from the IF statistical data and a total traffic amount for each flow calculated from the flow statistical data.