Virtual Network Controller Latency Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In cloud data centers, determining latency in a physical network is challenging due to the lack of readily available information about packet travel times and paths within virtual network architectures, which can lead to undetected bottlenecks and network performance issues.
Innovation Solution
A method involving a virtual network controller that collects and analyzes packet signatures, timestamps, and device identifiers from network devices to determine latency, identifying slow network devices and bottlenecks without the need for extensive testing, using packet classifiers to specify monitored packets and restrict data collection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If packet monitoring and latency measurement are implemented in virtualized networks, then network performance visibility and bottleneck detection are improved, but system complexity and resource overhead increase
Solution Approach 1:
The patent introduces an analytics engine as an intermediary component that collects packet monitoring data from multiple network devices, processes the information centrally, and generates latency measurements. This mediator approach allows distributed data collection without requiring complex latency calculation logic in each network device, thereby improving measurement precision while managing system complexity through centralized processing.
2Loss of information
If packet copies are sent to analytics engine for analysis, then latency information becomes available, but network bandwidth is consumed and packet processing overhead increases
Solution Approach 1:
The patent extracts only the essential latency-critical information from packets (such as timestamps and packet identifiers) rather than transmitting complete packet copies to the analytics engine. This extraction approach provides sufficient latency measurement data while significantly reducing the bandwidth consumption and processing overhead associated with transmitting full packet content.
3Measurement precision
If comprehensive packet monitoring is performed across all network devices, then complete latency data is obtained, but processing time and computational resources increase
Solution Approach 1:
The patent implements partial monitoring by selecting representative packets for latency analysis rather than processing every packet across all network devices. The analytics engine monitors a subset of packets that provide sufficient statistical information about network latency, achieving acceptable measurement precision while significantly reducing the computational burden and processing time required for comprehensive analysis.
Data Source
AI summary
Techniques are described for determining latency in a physical network that includes a number of network devices over which packets travel. A virtual network controller receives a plurality of messages from a plurality of network devices in a network, each of the messages including a packet signature comprising a hash of an invariant portion of an original packet that uniquely identifies the original packet, an identifier of one of the plurality of network devices from which the respective message was received, and a timestamp indicating a time an original packet was processed by the network device from which the respective message was received. The virtual network controller determines a latency of a physical network path in the network based on analysis of contents of the identified messages having a common packet signature.