SDN Performance Modeler for Network Impairment Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Testing software defined networks (SDNs) is challenging due to their ability to change rapidly, making it difficult for network operators to predict and verify network performance and quality of service, especially in software defined wide area networks (SD-WANs).
Innovation Solution
A system comprising a traffic generator, network impairment module, and network testing monitor that applies impairment profiles to SDN test traffic, maps these impairments to performance metrics, and uses an SDN performance modeler to train machine-learning models for predicting network performance under various conditions, including control plane and user plane traffic simulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If network testing systems are used to test SDNs, then network performance can be monitored and analyzed, but the rapid changes in SDN make it difficult to predict and verify network performance and quality of service
Solution Approach 1:
The system applies impairment profiles to test traffic before actual network operation to predict performance outcomes. By proactively introducing controlled impairments and measuring their effects, the system establishes performance baselines and predictions before the SDN undergoes rapid changes, enabling operators to verify quality of service in advance.
Solution Approach 2:
The network testing monitor continuously monitors SDN test traffic and feeds performance data back to the SDN performance modeler. This feedback loop enables the system to adapt to rapid SDN changes by continuously updating performance models and predictions, maintaining reliable performance verification despite dynamic network conditions.
2Measurement precision
If impairment profiles are applied to test traffic, then network performance under various conditions can be predicted, but the complexity of systematically testing multiple impairment scenarios increases
Solution Approach 1:
The SDN performance modeler serves multiple functions: it maps impairment profiles to performance metrics, trains machine learning models, and predicts network performance under various conditions. By consolidating these diverse functions into a single multi-functional component, the system achieves precise measurement capabilities without proportionally increasing overall system complexity.
Solution Approach 2:
The system uses machine learning models to create virtual copies of network performance behavior. Instead of physically testing every possible impairment scenario, the trained models replicate performance outcomes, allowing precise prediction of network behavior under multiple impairment conditions without proportionally increasing testing infrastructure complexity.
3Reliability
If machine learning models are trained on impairment data, then network issues can be predicted, but the time and resources required for systematic testing and model training increase
Solution Approach 1:
The system trains machine learning models in advance using impairment profile data before actual network deployment. By performing this training action preliminarily, the system establishes predictive capabilities ahead of time, reducing the time needed for real-time network issue prediction during actual operation.
Solution Approach 2:
The performance modeler continuously monitors SDN test traffic and continuously trains and updates machine learning models with new data. This continuous action allows the system to maintain accurate prediction capabilities over time, spreading the training workload continuously rather than requiring intensive batch training that would cause significant time loss.
Data Source
AI summary
Methods, systems, and computer readable media for testing software defined networks (SDNs). An example system includes a traffic generator configured for generating SDN test traffic for the SDN; a network impairment module configured for applying a plurality of impairment profiles to the SDN test traffic; and a network testing monitor configured for monitoring the SDN test traffic on the SDN. The system includes an SDN performance modeler configured for mapping, for each impairment profile, the impairment profile to SDN performance metric values based on monitoring the SDN test traffic by the network testing monitor in response to applying the impairment profile by the network impairment module.