Fault-Tolerant Routing for Toroidal Networks With Sparse Link Failures

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional dimension-order routing strategies in toroidal network topologies fail to guarantee node-to-node reachability and result in high load imbalance and traffic congestion due to sparse link failures, especially in supercomputers with centralized circuit switches.

Innovation Solution

A fault-tolerant routing algorithm that generates multiple fault-free routes by taking arbitrary hops from either the source or destination node, combined with dimension-order hops, and uses integer linear programming for load balancing to minimize link loads.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional dimension-order routing is used, then routing simplicity is maintained, but node-to-node reachability cannot be guaranteed due to link failures

Engineering Contradiction:

Improvenode-to-node reachabilityVSAvoidrouting algorithm complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The routing algorithm dynamically adapts to link failures by detecting faulty links and switching from conventional dimension-order routing to fault-tolerant routing modes. The system maintains multiple routing options and selects appropriate paths based on real-time link status, ensuring reachability while managing complexity through adaptive behavior rather than static complex routing tables.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The routing strategy changes parameters such as dimension ordering and hop selection based on detected link failures. When faults are detected, the algorithm modifies routing parameters to avoid faulty links while maintaining toroidal topology constraints, thereby guaranteeing reachability without requiring completely complex alternative routing infrastructure.

Inventive Principle:

Principle #35Parameter changes

2Ease of operation

If conventional dimension-order routing is used, then routing implementation is simple, but load imbalance and traffic congestion occur at links near circuit switches

Engineering Contradiction:

Improverouting implementation easeVSAvoidnetwork throughput

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The routing system dynamically adjusts path selection to balance load across the network. By detecting congestion patterns and link utilization levels, the algorithm adaptively redirects traffic away from overloaded links near circuit switches, maintaining simple implementation while improving throughput through runtime parameter adjustments rather than complex static load balancing.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The routing algorithm incorporates feedback mechanisms that monitor link utilization and traffic patterns. Based on this feedback, the system adjusts routing decisions to distribute load more evenly across the toroidal network, preventing congestion at critical links while maintaining implementation simplicity through localized adaptive responses rather than global complex control.

Inventive Principle:

Principle #23Feedback

3Reliability

If multiple fault-free routes are generated, then fault tolerance is improved, but routing computation time increases

Engineering Contradiction:

Improvefault toleranceVSAvoidrouting computation time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-computing and caching fault-free routes for common source-destination pairs. When link failures are detected, the algorithm quickly retrieves pre-computed alternative routes rather than performing full pathfinding computations, thereby maintaining high fault tolerance while minimizing real-time computation time and overhead.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The routing algorithm generates a limited set of candidate fault-free routes rather than exhaustively computing all possible paths. By focusing on a sufficient subset of alternative routes that meet fault tolerance requirements, the system achieves adequate reliability without the excessive computation time that would result from generating and evaluating all possible fault-free paths.

Inventive Principle:

Principle #16Partial or excessive action

4Productivity

If load balancing optimization is performed, then link load distribution is improved, but optimization processing time increases

Engineering Contradiction:

Improvenetwork efficiencyVSAvoidoptimization processing time

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The load balancing algorithm optimizes routing by adjusting parameters such as dimension order, hop selection, and path preference based on current network conditions. By changing these parameters dynamically rather than performing complete re-optimization, the system improves link load distribution while keeping processing time manageable through localized parameter adjustments rather than global optimization computations.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12418475B2Fault-tolerant routing algorithm for toroidal network topologies

Publication Date: 2025.09.16 GOOGLE LLC

AI summary

Generally disclosed herein is an approach for optimizing routing strategy to tolerate faults in a toroidal network topology including, but not limited to, N-dimensional mesh, torus, and twisted torus. The approach may include balancing a load for a specified input traffic pattern operating offline or online. The approach may also include an optimization enhancement technique specifically applicable to symmetric, dynamically composable toroidal networks based on a set of centrally connected circuit switches.