Predictive Congestion Detection in High Radix Routers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In high radix routers, the latency caused by stale congestion state information significantly reduces network throughput, as adaptive routing decisions are made based on outdated information, leading to suboptimal routing choices.
Innovation Solution
Implementing stateful routing engines that track usage information and make predictions about congestion changes between updates, allowing for more informed routing decisions and reducing the impact of stale congestion information by altering congestion values based on traffic patterns and timing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If adaptive routing decisions are made based on congestion information from outputs, then routing flexibility is improved, but the information becomes stale due to feedback latency
Solution Approach 1:
The patent applies preliminary action by having input routing engines predict congestion states at outputs before the actual congestion feedback arrives. The routing engine uses known usage information (traffic patterns, historical data) to proactively estimate future congestion states, allowing routing decisions to be based on predicted rather than stale information. This resolves the contradiction by maintaining routing flexibility while eliminating the information staleness problem.
2Loss of information
If congestion feedback latency is reduced by minimizing outputs, then information freshness is improved, but routing flexibility is reduced
Solution Approach 1:
The patent introduces an intermediary prediction mechanism that mediates between the need for fresh congestion information and routing flexibility. Instead of directly reducing the number of outputs (which would lose flexibility), the system uses prediction algorithms as an intermediary layer that generates fresh congestion estimates for all outputs based on usage information. This allows the system to maintain full routing flexibility while obtaining fresh congestion information through prediction rather than direct feedback.
3Productivity
If routing decisions use current congestion information, then routing optimality is improved, but feedback latency increases
Solution Approach 1:
The patent applies preliminary action by performing congestion prediction in advance of when congestion feedback would normally be received. Input routing engines continuously update predictions using usage information before routing decisions are required, ensuring that optimal routing decisions can be made immediately without waiting for latency-prone feedback loops. This resolves the contradiction by enabling current/optimal routing decisions through predictive action rather than reactive feedback.
Data Source
AI summary
A system and method for predictive congestion detection for network devices is provided. A routing engine associated with an input of a router receives congestion information from an output, utilizing the received congestion information to initialize a congestion value associated with that output. Between receipt of updated congestion information from the output, the routing engine predicts a potential change in the congestion state at the output based on the congestion value and information regarding usage of the output that is known to the routing engine.