Traffic Flow Classification for Network Appliance Data Path Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network appliances such as switches and routers face challenges in adapting to changes in feature sets, protocols, operating systems, and hardware configurations, leading to inefficiencies in processing high volumes of packets and limited flexibility in data plane operations.
Innovation Solution
Implementing a method in network appliances that uses a classification model to predict whether traffic flows are long lived or short lived, directing them through either a fast data path or a slow data path based on these predictions, enhancing hardware resource utilization and throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If network appliances process all traffic flows through the same data path, then simplicity of data plane operations is maintained, but hardware resource utilization efficiency deteriorates
Solution Approach 1:
The patent divides the data plane into two separate data paths: a fast data path for long-lived flows and a slow data path for short-lived flows. This segmentation allows each path to be optimized for its specific traffic type, improving hardware resource utilization efficiency while maintaining manageable complexity through clear functional separation.
Solution Approach 2:
The patent applies different processing qualities to different traffic flows by routing long-lived flows through the fast data path with optimized hardware resources and short-lived flows through the slow data path. This local quality differentiation ensures that each traffic type receives appropriate processing resources, improving overall hardware utilization efficiency.
2Productivity
If network appliances prioritize long lived flows in the fast data path, then throughput and power efficiency are improved, but complexity of traffic flow classification increases
Solution Approach 1:
The patent performs preliminary classification of traffic flows into long-lived and short-lived categories before routing them to appropriate data paths. By predicting flow characteristics in advance using machine learning models, the system can prioritize long-lived flows in the fast data path, improving throughput and power efficiency without adding complex real-time classification requirements during packet processing.
Solution Approach 2:
The patent introduces a machine learning-based flow predictor as an intermediary component that analyzes traffic flow characteristics and determines the appropriate data path. This intermediary handles the complexity of traffic flow classification, allowing the fast and slow data paths to operate with simplified logic while still achieving intelligent routing for improved throughput and power efficiency.
3Adaptability or versatility
If network appliances use machine learning models for traffic flow classification, then adaptability to changing traffic patterns is improved, but computational overhead and processing time increase
Solution Approach 1:
The patent performs machine learning-based traffic flow classification in advance, before packets enter the fast or slow data path. By completing the computationally intensive classification process beforehand, the system achieves high adaptability to changing traffic patterns while minimizing the time loss during actual packet processing, as the routing decisions are already determined.
Solution Approach 2:
The patent replaces traditional rule-based traffic classification mechanisms with machine learning models that can adapt to changing traffic patterns. This substitution improves adaptability while the preliminary action approach ensures that the computational overhead does not significantly impact packet processing time, as the classification is completed before the time-critical forwarding decision.
Data Source
AI summary
Methods and system for directing traffic flows to a fast data path or a slow data path are disclosed. Parsers can produce packet header vectors (PHVs) for use in match-action units. The PHVs are also used to generate feature vectors for the traffic flows. A flow training engine produces a classification model. Feature vectors input to the classification model result in output predictions predicting if a traffic flow will be long lived or short lived. The classification models are used by network appliances to install traffic flows into fast data paths or the slow data paths based on the predictions.