Distributed HSRP Gateways in VxLAN Flood-and-Learn Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current HSRP implementations in VxLAN environments are limited to supporting only two active gateways, which restricts scalability and redundancy in data center networks.
Innovation Solution
The method involves configuring HSRP on the spine layer or leaf switches, using a special VTEP address for all HSRP peers, and leveraging underlying routing protocols like ISIS or OSPF to elect active, standby, and listen nodes based on priority, allowing for N active distributed gateways by load-sharing unicast traffic and fast convergence in case of peer down detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If HSRP is deployed in traditional configurations (vPC pair or single active), then gateway redundancy is provided, but the number of active gateways is limited to two
Solution Approach 1:
The patent segments the traditional single HSRP group into multiple HSRP groups (Group 1, Group 2, etc.), each with its own virtual IP address and virtual MAC address. This allows different sets of gateway devices to serve different VLANs or VxLAN segments, enabling multiple active gateways simultaneously while maintaining HSRP redundancy benefits. Each segment operates independently, breaking the limitation of having only one active gateway per HSRP group.
Solution Approach 2:
The patent introduces a new dimension by deploying HSRP gateways across multiple layers (spine layer and leaf switch layer) rather than confining them to a single layer. This multi-layer deployment creates additional active gateway instances distributed throughout the network fabric, transforming the traditional single-plane HSRP architecture into a multi-dimensional gateway distribution model that supports N active gateways.
2Reliability
If HSRP is configured on spine layer switches, then routing between VxLAN segments is improved, but convergence speed during failover is slowed
Solution Approach 1:
The patent merges the control plane functions of multiple HSRP gateway instances into a coordinated system where spine layer switches and leaf switch layer gateways work together. The HSRP hello messages and state synchronization mechanisms are enhanced to enable faster detection of gateway failures and quicker failover decisions, reducing convergence time while maintaining the routing capabilities provided by spine layer deployment.
Solution Approach 2:
The patent implements preliminary actions by pre-configuring multiple HSRP gateway instances with predetermined priorities and roles before failover events occur. The HSRP protocol parameters, including hello intervals and hold timers, are optimized in advance to enable rapid convergence. Standby gateways are pre-positioned and pre-synchronized, so when an active gateway fails, the transition to a new active gateway occurs quickly without extensive recalculation or reconfiguration.
Data Source
AI summary
A method of supporting N active, distributed HSRP gateways in a virtual Extensible local area network. The method includes: joining by HSRP active, HSRP standby, HSRP listen gateways a special virtual Extensible LAN tunnel endpoint address; advertising by HSRP active, HSRP standby, HSRP listen gateways its address; assigning priority to HSRP active, HSRP standby, HSRP listen gateways; determining the liveliness of the HSRP active gateway; and the HSRP active gateway sourcing Hello with HSRP VMAC in the inner layer 2 header, and the special virtual Extensible LAN tunnel endpoint address in the outer Layer 3 address; replying by the HSRP active gateway to a broadcast ARP to an HSRP Virtual IP address; forwarding packets destined to Virtual MAC by any of HSRP active, HSRP standby, HSRP listen gateways and replying by HSRP active, HSRP standby, HSRP listen gateways to a unicast ARP to an HSRP Virtual IP address.