Conversational Learning for Scalable Data Center Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data center networks face challenges in optimizing resource utilization and ensuring resilient infrastructure to support diverse applications and services, particularly in massively scalable environments where current solutions lead to sub-optimal traffic routing and increased complexity due to limited FIB table space and scalability issues.
Innovation Solution
Implementing conversational learning in network environments by using a system and method that installs subnet entries with glean adjacencies in FIB/ADJ tables, allowing on-demand installation of remote host routes based on active conversations, and leveraging extended community attributes for routing decisions, thereby optimizing traffic routing and minimizing manual configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional routing methods are used in data center networks, then all host routes are installed in FIB tables, but FIB table space is exhausted and device complexity increases in massively scalable environments
Solution Approach 1:
The patent segments routing information into two types: subnet routes (installed in hardware FIB tables) and host routes (installed in software FIB tables). This segmentation allows the system to maintain complete routing information while distributing the storage burden between hardware and software, resolving the contradiction between routing completeness and FIB table space constraints.
Solution Approach 2:
The patent implements dynamic route installation where host routes are installed in software FIB tables on-demand based on active conversations. Routes are dynamically added when needed and removed when no longer required, allowing the system to adapt to changing traffic patterns and maintain routing completeness without permanently consuming hardware FIB table space.
2Measurement precision
If all host routes are installed in FIB tables to ensure complete routing, then routing accuracy is maintained, but traffic routing optimization is reduced due to sub-optimal paths
Solution Approach 1:
The patent introduces a software FIB table as an intermediary layer between the hardware FIB table and the routing decision process. The software FIB table stores host routes that are not present in hardware, acting as a mediator that enables accurate routing decisions for all destinations while allowing the hardware FIB table to maintain optimized subnet routes for high-speed forwarding.
3Measurement precision
If manual configuration is used for routing, then routing precision is maintained, but network management complexity increases
Solution Approach 1:
The patent implements automated route installation and removal based on active conversation detection. The system automatically monitors traffic patterns, installs host routes in software FIB tables when conversations are detected, and removes them when no longer needed, eliminating the need for manual configuration while maintaining routing precision through attribute-based routing decisions.
Data Source
AI summary
A system and a method for providing conversational learning is implemented in a network environment. An exemplary method includes receiving a subnet route advertisement that includes an attribute that triggers glean behavior for routing decisions; and installing a subnet entry in a Forwarding Information Base/Adjacency (FIB/ADJ) table. The subnet entry includes a subnet associated with the subnet route advertisement and a corresponding glean adjacency. The corresponding glean adjacency is configured to trigger installation of a host entry associated with a host in an active conversation in a network.