Hybrid VRF Labels for ECMP Forwarding Loop Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional VRF VPN label allocation in multi-homed provider edge nodes with equal-cost multipath routing (ECMP) constrains network topologies and inefficiently uses storage space, particularly in Layer-3 VPNs, leading to undesired forwarding behaviors and slow path failure recovery.
Innovation Solution
Implementing per-VRF and per-CE labels on egress nodes, with hybrid VRF labels that include both per-VRF and per-CE labels, to configure forwarding information and control plane advertisements, allowing for efficient ECMP FEC table construction and avoiding forwarding loops during path failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional VRF VPN labels are allocated for ECMP with multi-homed VRFs, then forwarding of VRF traffic to appropriate CE nodes is enabled, but network topology is constrained and storage space on PE hardware is consumed
Solution Approach 1:
The patent segments the ECMP FEC entry structure by separating VRF-specific forwarding information from ECMP path selection information. This is achieved by introducing VRF-specific next-hop identifiers and outgoing interface identifiers that are stored separately from the shared ECMP FEC entries, thereby reducing the storage burden on PE hardware while maintaining full forwarding capability.
Solution Approach 2:
The patent adds a new dimension to the forwarding architecture by introducing a hierarchical label structure with VRF-specific labels and ECMP-specific labels. This multi-dimensional approach allows the system to maintain detailed VRF forwarding information without proportionally increasing ECMP FEC entry storage, as the VRF-specific information is cached separately rather than replicated in each ECMP entry.
2Ease of operation
If conventional VRF VPN labels are used for ECMP scenarios, then VRF traffic forwarding is supported, but network topology flexibility is constrained
Solution Approach 1:
The patent creates a universal forwarding mechanism that works across diverse network topologies by introducing abstract next-hop identifiers and outgoing interface identifiers that can represent any combination of physical and virtual paths. This universal approach allows the same ECMP FEC entry structure to support various topologies including multi-homed VRFs, standard ECMP, and hierarchical routing scenarios without requiring topology-specific configurations.
Solution Approach 2:
The patent introduces dynamic path selection capabilities by allowing the best path determination to consider multiple factors including VRF-specific metrics, ECMP cost values, and real-time path status. The system can dynamically adjust which next-hop is selected as best based on changing network conditions, providing adaptability to various topology configurations and failure scenarios.
3Reliability
If per-VRF and per-CE labels are implemented on egress nodes, then forwarding loops are prevented and path failure recovery is enabled, but forwarding information complexity increases
Solution Approach 1:
The patent implements preliminary action by pre-configuring backup next-hops and outgoing interfaces for each VRF FEC entry before any path failure occurs. The best path is predetermined based on ECMP cost values and VRF-specific metrics, and backup paths are pre-identified, enabling immediate failover without complex real-time decision-making when failures occur.
Solution Approach 2:
The patent implements feedback mechanisms by monitoring the status of next-hops and outgoing interfaces, and using this information to dynamically adjust path selection. The system continuously evaluates path health and can switch from primary to backup paths based on real-time feedback about path availability, ensuring reliable forwarding while maintaining manageable complexity through automated status monitoring.
Data Source
AI summary
First and second egress nodes are each multi-homed to a customer edge (CE) that participates in virtual routing and forwarding (VRF). First forwarding information is configured on the first egress node. The first information includes VRF labels and defines forwarding of traffic based on the VRF labels and a status of a primary path to the CE. The VRF labels include a per-VRF label for the VRF and a per-CE label for the CE. Second forwarding information is configured on the second egress node. The second forwarding information includes the per-VRF label and the per-CE label, and defines traffic forwarding based on the VRF labels. Upon receiving traffic for the CE that carries the per-VRF label, the first egress node determines the status of the primary path, and forwards the traffic to either the CE over the primary path or to the second egress node, depending on the status.


