SDN Reachability Graph for Network Failure Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional network systems consume excessive resources and bandwidth to detect node and link failures, as they rely on hello packets to check end-to-end connectivity, leading to inefficient failure detection and potential data loss.
Innovation Solution
Implementing a software-defined networking (SDN) infrastructure with status agents installed on nodes and switches that periodically gather and send status information to an overlay network controller, generating a reachability graph to quickly identify and bypass failed nodes or links, reducing the need for conventional hello packets and minimizing resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional hello packets are used to check end-to-end connectivity between nodes, then node and link failure detection can be achieved, but excessive bandwidth and CPU resources are consumed
Solution Approach 1:
The patent introduces an SDN controller as an intermediary that centralizes the connectivity check function. Instead of each node pair exchanging hello packets directly, the SDN controller receives status information from all nodes and switches, performs centralized reachability analysis using a reachability graph, and determines connectivity status. This intermediary approach eliminates the need for numerous node-to-node hello packets, significantly reducing bandwidth and CPU resource consumption while maintaining reliable failure detection capability.
2Reliability
If hello packets are exchanged at application level between all application pairs, then connectivity status can be monitored, but unnecessary traffic bandwidth and CPU processing are required
Solution Approach 1:
The patent extracts the connectivity monitoring function from the application layer and moves it to the SDN control plane. Application status information is collected by status agents at the network infrastructure layer, and the SDN controller performs centralized reachability analysis. This extraction eliminates the need for application-level hello packet exchanges, reducing traffic bandwidth and CPU processing overhead by orders of magnitude while maintaining comprehensive connectivity monitoring across all application pairs.
Solution Approach 2:
The patent merges the connectivity monitoring function for all application pairs into a single centralized SDN controller. Instead of each application pair independently exchanging hello packets, the SDN controller consolidates status information from all nodes and switches, builds a global reachability graph, and determines connectivity for all application pairs simultaneously. This merging approach dramatically reduces system complexity and resource consumption.
3Reliability
If conventional network elements monitor connectivity independently, then node aliveness can be checked, but the total time and resource consumption is directly proportional to the number of nodes and links
Solution Approach 1:
The patent segments the connectivity monitoring function into two distinct layers: (1) Status agents at each node and switch that collect local status information and send it to the SDN controller, and (2) The SDN controller that performs centralized reachability analysis using a global reachability graph. This segmentation allows parallel information collection at the data plane while centralized computation occurs at the control plane, significantly reducing the time and resources required for connectivity checks compared to independent monitoring by each network element.
Data Source
AI summary
Embodiments of the present invention include systems and methods for detecting status of network elements, such as, applications, information handling systems, links, and the like, in a network. In embodiments, a system for detecting status of network elements in a network includes an agent that monitors status of one or more network elements related to a first information handling systems and sends a status notice to a status controller. In embodiments, the status controller receives status information from a number of status agents operating on other information handling systems in the network and uses that information to generate one or more reachability graphs. In embodiments, the status controller sends at least part of a reachability graph to one or more of the status agents in the network. In embodiments, an application on an information handling systems may query the local status agent to obtain status information regarding a network element.