Distributed IPv6 Neighbor Discovery for Switch Scalability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As the number of virtual machines and hosts grows, traditional network architectures face scalability issues due to overwhelming neighbor discovery messages exchanged between hosts and central controller processors, leading to potential overload and failures.
Innovation Solution
Implementing a distributed IPv6 neighbor discovery module that distributes neighbor discovery traffic across multiple switching devices, using IPv6 address formats to configure port and device prefixes in FIB CAM tables, and employing access lists to punt specific neighbor discovery messages to switch device processors, while routing other traffic through the central controller.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all neighbor discovery messages are processed by a central controller processor, then centralized control and management are maintained, but the central controller becomes overloaded and system reliability decreases as the number of hosts grows
Solution Approach 1:
The patent divides the neighbor discovery processing function into two segments: (1) centralized controller that handles initial neighbor discovery protocol operations and maintains the neighbor discovery cache, and (2) distributed switch device processors that independently process neighbor discovery messages using the cache. This segmentation distributes the processing load from the central controller to edge devices, improving reliability while maintaining centralized coordination.
Solution Approach 2:
The patent introduces a neighbor discovery cache as an intermediary data structure stored in FIB CAM tables on switch devices. This cache acts as a mediator between the central controller and neighbor discovery messages, allowing switch devices to independently process messages by querying the cache without constantly communicating with the central controller, thereby reducing controller load while maintaining centralized cache management.
2Adaptability or versatility
If the number of virtual machines and hosts increases, then network capacity and functionality improve, but neighbor discovery message volume overwhelms the central controller
Solution Approach 1:
The patent segments neighbor discovery processing so that switch device processors handle message processing independently using local FIB CAM tables, while the central controller only performs cache management. This allows the system to scale to support more hosts without proportionally increasing controller processing load, as the bulk of message handling occurs at the distributed switch level.
Solution Approach 2:
The patent implements partial processing at the switch device level by querying the FIB CAM table for neighbor discovery messages. When a match is found in the cache, the switch device processes the message independently without involving the central controller. This partial action at the edge reduces the volume of messages requiring central controller processing, enabling the system to handle increased host capacity while maintaining controller productivity.
3Productivity
If neighbor discovery messages are distributed to multiple switching devices, then processing load is reduced and scalability improves, but coordination and message routing complexity increases
Solution Approach 1:
The patent merges the neighbor discovery cache with the FIB CAM table structure on switch devices. By combining these data structures and using the existing FIB CAM infrastructure for storage and retrieval, the patent reduces the complexity of distributed coordination. The unified structure allows switch devices to efficiently query neighbor discovery information using the same mechanisms already in place for forwarding information, avoiding the need for separate distributed cache management protocols.
Solution Approach 2:
The FIB CAM table structure serves multiple functions: it stores forwarding information, maintains the neighbor discovery cache, and enables both normal packet forwarding and neighbor discovery message processing. This multi-functionality reduces coordination complexity by using a single unified data structure and processing mechanism for multiple purposes, rather than requiring separate specialized structures for each function.
Data Source
AI summary
A distributed neighbor discovery module is disclosed where all neighbor discovery operations and functionalities may be distributed to a switch device processor. Each neighbor discovery process on a switch device may run as an individual neighbor discovery router node. There may be no coordination between neighbor discovery processes occurring on different switch devices. All other traffic, including both IPv6 control traffic and data traffic destined to the router may be processed by the central controller processor.