Peer Link-Less MC-LAG Using Spine-Level Sub-LAG Forwarding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current MC-LAG network deployments with peer links in data centers face limitations such as reduced bandwidth, increased costs, and scaling issues due to the use of inter-node links between peer information handling system nodes.
Innovation Solution
Implementing sub-LAGs (link aggregation groups) managed by spine nodes to handle data traffic, eliminating the need for inter-node links by using spine nodes to determine port connections based on learned connectivity information and forming sub-LAGs to reach dual-homed, single-homed, or orphaned nodes, with indicators for broadcast, unknown unicast, and multicast traffic.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If inter-node links are used to connect peer information handling system nodes in MC-LAG deployments, then multi-homing connectivity is achieved, but bandwidth is reduced and costs increase
Solution Approach 1:
The patent extracts and eliminates the inter-node link (peer link) from the MC-LAG architecture. By removing this redundant connection between peer nodes, the system achieves the same multi-homing connectivity functionality without consuming bandwidth on dedicated peer links, thereby increasing available bandwidth for actual data traffic while maintaining reliability through alternative paths via upstream switches.
2Reliability
If inter-node links are used to connect peer information handling system nodes in MC-LAG deployments, then multi-homing connectivity is achieved, but costs increase
Solution Approach 1:
The patent removes the inter-node link component from the MC-LAG deployment, eliminating the hardware and configuration costs associated with implementing and maintaining peer links between information handling system nodes, while preserving multi-homing connectivity through upstream switch-based path management.
3Reliability
If inter-node links are used in MC-LAG deployments, then peer node communication is enabled, but scaling is limited
Solution Approach 1:
The patent implements a scalable architecture where upstream switches perform universal path management and connectivity functions for all peer nodes in the network. This centralized approach allows the system to scale to multiple peer nodes without requiring dedicated inter-node links between each pair, enabling the network to grow while maintaining peer node communication capabilities through the shared upstream switch infrastructure.
Data Source
AI summary
Current data center deployments, particularly layer 2 (L2) deployments that use multi-chassis link aggregation group (MC-LAG) implementations, have limitations. Such deployments use a link, known as an inter-node link (INL)/inter-chassis link (ICL), to connect peer information handling system nodes. Because the peer nodes are coupled together via one or more INL connections, there are fewer ports available for each peer node to connect to end nodes. This configuration creates limitations to bandwidth and scaling, and also increases costs. Embodiments herein allow for the elimination of INLs by moving forwarding decisions to the spine node level. In one or more embodiments, a spine node determines ports to reach dual-homed, single-homed, and orphaned nodes that are connected to leaf node(s) based on information learned from the leaf nodes. For a leaf node, a sub-LAG (link aggregation group) may be created to reach single-homed/orphaned nodes connected to the leaf node.