Segment Routing VPN Signaling for Fast MAC Learning Convergence

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing EVPN, NG-VPLS, VPWS, and L3VPN signaling technologies face scalability, convergence, and overhead challenges due to reliance on BGP control plane MAC learning and significant BGP overhead, leading to slow reaction times and increased operational complexity.

Innovation Solution

Implementing Segment Routing to enable data plane MAC learning and reduce BGP overhead by using Segment Identifiers (SIDs) for EVPN, NG-VPLS, and L3VPN signaling, enabling fast convergence, auto-discovery, and active/active multihoming without the need for EVPN Route Types 1, 2, 3, and 4, and reducing control plane overhead by at least two orders of magnitude.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If BGP control plane MAC learning is used, then MAC address distribution is achieved, but convergence speed is slow and reaction time to MAC moves is delayed

Engineering Contradiction:
Improveconvergence speedVSAvoidreaction time to MAC moves
Core Design Contradiction:
SpeedVSLoss of time

Solution Approach 1:

The patent replaces the control plane BGP-based MAC learning mechanism with a data plane MAC learning mechanism. Instead of using BGP protocol messages to learn and distribute MAC addresses, the system uses actual data traffic flow to learn MAC addresses dynamically, achieving faster convergence and immediate reaction to MAC moves without the delays inherent in control plane processing.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Quantity of substance

If BGP EVPN signaling is used for E-tree services, then MAC address learning is achieved, but control plane overhead and scalability are significantly impacted

Engineering Contradiction:
ImproveMAC address learning capabilityVSAvoidcontrol plane overhead
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The patent extracts the MAC address learning function from the BGP control plane and relocates it to the data plane. This separation removes the burden of MAC address distribution from the control plane, significantly reducing control plane overhead and improving scalability while maintaining the essential capability to learn and forward MAC addresses through data traffic.

Inventive Principle:
Principle #2Taking out (Extraction)

3Adaptability or versatility

If traditional VPLS with Pseudowires is used, then L2 Ethernet service is provided, but scalability becomes a bottleneck with large scale deployment

Engineering Contradiction:
ImproveL2 Ethernet service capabilityVSAvoidPW maintenance overhead
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges multiple Pseudowire connections into a single EVPN instance, allowing multiple L2 Ethernet services to be provided through a unified control plane framework. This consolidation eliminates the need to individually manage and maintain numerous PW connections, significantly improving scalability while preserving full L2 Ethernet service functionality.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP4221138B1EVPN signaling, NG-VPLS e-tree signaling, VPWS signaling, and l3VPN signaling using segment routing or mpls
Publication Date: 2026.04.15 CIENA CORP
  • EP4221138B1 patent drawingFigure 1
  • EP4221138B1 patent drawingFigure 2
  • EP4221138B1 patent drawingFigure 3

AI summary

A method (350) comprising steps of: determining (352) a plurality of services supported at a node; determining (354) a bitmap to represent the plurality of services supported at the node, wherein the bitmap includes a starting service and each subsequent bit representing another service of the plurality of services and with each bit in the bitmap set based on the plurality of services supported at the node; and transmitting (356) an advertisement to nodes in a Segment Routing network (10) with the bitmap based on the plurality of services supported at the node.