Flow-Based Latency Measurement in Overlay Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Software-Defined Networking (SDN) environments, conventional methods are inefficient in diagnosing and addressing latency-related issues in logical overlay network traffic, as they rely on indirect indicators and lack real-time responsiveness, especially when measuring overall latency between hosts, which does not account for varying physical forwarding paths and congestion affecting individual packet flows.
Innovation Solution
Implementing flow-based latency measurement by generating and sending encapsulated packets with timestamp information between endpoints over logical overlay tunnels, allowing for end-to-end latency calculation at the granularity of a packet flow without requiring clock synchronization between hosts, thereby facilitating precise diagnosis and troubleshooting of network performance issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional latency measurement methods are used in SDN environments, then overall latency between hosts can be measured, but the measurement lacks precision for individual packet flows and does not account for varying physical forwarding paths
Solution Approach 1:
The patent segments the network latency measurement into individual packet flow levels rather than measuring only overall host-to-host latency. By breaking down the measurement into per-flow granularity, the system can identify specific latency issues affecting individual traffic streams while maintaining manageable complexity through structured implementation
Solution Approach 2:
The patent introduces intermediary components including virtual tunnel endpoints (VTEPs) and flow counters that act as mediators between physical network infrastructure and virtual machine traffic. These intermediaries enable precise flow-based latency measurement by capturing timestamp information at strategic points without requiring complex modifications to end systems
2Reliability
If flow-based latency measurement is implemented with per-flow granularity, then accurate diagnosis of network performance issues is enabled, but the complexity of implementing and managing the measurement system increases
Solution Approach 1:
The patent implements universal flow counter components that can handle multiple packet flows simultaneously through a single standardized interface. The flow counter structure is designed to be multi-functional, capable of tracking various traffic types and protocols through consistent mechanisms, thereby improving diagnosis reliability without proportionally increasing system complexity
Solution Approach 2:
The measurement system implements self-service capabilities through automatic timestamp generation and flow identification. The virtual tunnel endpoints automatically generate flow-based identifiers and capture timing information without requiring manual configuration for each flow, enabling reliable per-flow measurement while keeping operational complexity manageable
Data Source
AI summary
Example methods and systems for flow-based latency measurement for logical overlay network traffic are described. In one example, in response to detecting a first inner data packet associated with a packet flow, a first computer system may generate and send a first encapsulated packet via a logical overlay tunnel towards a second computer system. The first encapsulated packet may be generated by encapsulating the first inner data packet with a first outer header that includes first time information associated with the first inner data packet at the first computer system. In response to detecting a second encapsulated packet from the second computer system via the logical overlay tunnel, the first computer system may determine a flow-based latency measurement associated with the packet flow based on the first time information, and second time information identified from a second outer header of the second encapsulated packet.