Forwarding Unit FIB Memory Optimization via Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing FIB models in data communication devices, such as unified and scatter table models, face issues with large memory usage and slow updating speeds, and fail to accommodate different private domains for prefixes, leading to inefficient forwarding and inability to perform distinct processes for different prefixes.
Innovation Solution
A method and apparatus that save Route Prefixes with private data and corresponding Next Hop Arrays, using a scatter table model with added private domains and flag bits to enable efficient forwarding and distinct processes for each prefix, reducing memory usage and improving updating speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a unified table model is used where each prefix corresponds to a Next Hop Array, then the FIB lookup can be performed efficiently, but the memory space occupied becomes very large because the same next hop is copied tens of thousands of times
Solution Approach 1:
The patent segments the FIB structure into two separate tables: a prefix table storing prefix information and a next hop table storing next hop information. This segmentation allows efficient lookup through the prefix table while avoiding the memory waste of copying next hop entries for every prefix, thus resolving the contradiction between lookup efficiency and memory usage.
Solution Approach 2:
The patent introduces a pointer structure as an intermediary between the prefix table and next hop table. The pointer stores only the address information of the next hop array, acting as a mediator that links prefixes to next hops without requiring actual copies of next hop data for each prefix, thereby reducing memory space while maintaining lookup efficiency.
2Ease of manufacture
If the same next hop has many copies in the Next Hop Array, then the FIB can be constructed simply, but the updating speed becomes low because all copies must be deleted when a next hop fails
Solution Approach 1:
By segmenting the FIB into separate prefix and next hop tables with a pointer linkage mechanism, the patent eliminates the need to delete multiple copies of next hop information. When a next hop fails, only the entry in the next hop table needs to be updated, dramatically improving updating speed while keeping the construction process simple.
Solution Approach 2:
The patent changes the copying approach from copying next hop data for each prefix to copying only the pointer address. This allows the FIB to be constructed simply by creating pointer references rather than full data copies, and enables fast updating by modifying only the pointer references in the prefix table rather than deleting multiple copies.
3Volume of stationary object
If a scatter table model is used with one copy of Next Hop Array, then memory space is reduced, but the ability to perform different processes for different prefixes is lost
Solution Approach 1:
The patent applies local quality by storing different private data for different prefixes in the prefix table while sharing the same next hop table structure. Each prefix can have its own private attributes (such as QoS parameters or routing policies) stored locally in the prefix table, enabling different processes for different prefixes without requiring separate next hop copies, thus maintaining adaptability while reducing memory usage.
Data Source
AI summary
A method for forwarding service in the data communication device and the forwarding apparatus, said apparatus includes a memory unit and a forwarding unit, wherein storing the Route Prefix, the private data of the Route Prefix and the corresponding Next Hop Array; said forwarding unit forwards the service at the data communication device according to said private data and the Next Hop corresponding with the Route Prefix. Using the present invention, it can implement the corresponding different a process for the different Prefix while the data communication device is forwarding service. It also may constitute a new-style FIB using the present invention, accordingly not only improve the updating speed of FIB, save the internal memory of the data communication device, but also satisfy the requirement that the different Prefixes need to have the different private domains with some specific functions.