IS-IS LFA Computation with Parallel Links
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current IS-IS routing protocols face inefficiencies in calculating loop-free alternative paths in networks with parallel links, leading to unnecessary computational costs and traffic disruptions due to the need for multiple SPF computations for each adjacency.
Innovation Solution
The proposed method reduces LFA SPF computations by storing equal and non-equal cost multi-path links in a neighbor data structure, allowing one LFA SPF computation from each neighbor node to consider all valid parallel adjacencies, thereby minimizing redundant calculations and improving computational efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple SPF computations are performed for each adjacency to calculate loop-free alternative paths in networks with parallel links, then the accuracy and completeness of LFA path identification is improved, but the computational cost and convergence time increase significantly
Solution Approach 1:
The patent merges multiple SPF computations into a single optimized computation by identifying and processing parallel links collectively. Instead of performing separate SPF computations for each adjacency, the system performs one SPF computation and then identifies parallel links among the resulting adjacencies, significantly reducing computational overhead while maintaining LFA identification accuracy.
Solution Approach 2:
The patent performs preliminary identification of parallel links immediately after a single SPF computation, before undertaking additional computations. By analyzing the adjacency information already obtained and identifying parallel links at this stage, the system avoids redundant computations and accelerates the LFA calculation process.
2Reliability
If multiple SPF computations are performed for each adjacency to ensure complete LFA path coverage, then the reliability of traffic disruption prevention is improved, but the computational resources consumed increase
Solution Approach 1:
The patent combines multiple SPF computation operations into a single computation followed by parallel link analysis. This merging approach maintains the reliability of LFA path identification by thoroughly analyzing all adjacencies from one comprehensive SPF run, while significantly reducing the computational resources that would be consumed by multiple separate SPF computations.
Solution Approach 2:
The system uses the adjacency information generated by the single SPF computation to self-identify parallel links and perform LFA calculations without requiring additional SPF computations. The preliminary adjacency data serves multiple purposes, enabling both primary routing determination and LFA identification through subsequent analysis of the same data structure.
3Measurement precision
If traditional SPF computation is used for each adjacency to calculate LFA paths, then the completeness of path analysis is improved, but the device complexity and processing overhead increase
Solution Approach 1:
The patent segments the LFA calculation process into two distinct phases: first performing a single comprehensive SPF computation to establish all adjacencies, then separately identifying parallel links among those adjacencies. This segmentation allows the system to maintain complete path analysis while avoiding the complexity of integrating multiple full SPF computations, as each phase handles a specific task with optimized processing.
Solution Approach 2:
The system performs preliminary SPF computation to establish the complete adjacency map before undertaking LFA-specific analysis. By having all adjacency information available from the preliminary SPF run, the system can efficiently identify parallel links and calculate LFAs without the overhead of repeated computations, maintaining analytical completeness while reducing processing complexity.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
A method is executed by a network device to a calculate loop free alternative (LFA) for each node in an intermediate system - intermediate system (IS-IS) area or IS-IS domain where the network device is connected with other nodes via multiple parallel links. The method includes a shortest path first (SPF) computation of the primary paths for the network device that tracks directly connected links from each node in the IS-IS area or IS-IS domain to the network device which is a source node for the SPF. These tracked links can then be utilized in subsequent LFA computations to avoid unnecessary calculations.