Unified Message Descriptor Linked List for Packet Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current packet-switched networks face complexity and high overhead in managing both unicast and multicast messages due to the need for separate management mechanisms, which increases maintenance costs and processing complexity.
Innovation Solution
A message processing method that allocates a single node in the cache space for both unicast and multicast messages, using a linked list to manage descriptor information, allowing for unified memory management and reducing the need for separate mechanisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate management mechanisms are used for unicast and multicast messages, then the messages can be managed according to their specific attributes, but the device complexity and maintenance overhead increase significantly
Solution Approach 1:
The patent merges the previously separate unicast and multicast management mechanisms into a single unified management structure. Both unicast and multicast messages are managed through the same cache space allocation and descriptor management processes, eliminating the need for duplicate management logic and reducing overall system complexity while maintaining the ability to handle different message types appropriately.
Solution Approach 2:
The management mechanism is designed to be universal, handling both unicast and multicast messages through the same procedures. The cache space allocation, descriptor generation, and queue management operations work for both message types without requiring separate specialized mechanisms, thereby reducing complexity while preserving functional correctness.
2Adaptability or versatility
If multiple nodes are allocated for multicast messages in the cache space, then multicast messages can be replicated and distributed to multiple queues, but the resource overhead and management complexity increase
Solution Approach 1:
Instead of allocating separate physical cache nodes for each multicast replica, the patent uses descriptor copying. A single cache node stores the original multicast message, and multiple descriptors pointing to this single node are created and distributed to different queues. This approach maintains multicast distribution capability while significantly reducing cache space consumption.
Solution Approach 2:
The descriptor acts as an intermediary between the cache space and the queues. Rather than directly allocating cache nodes for each queue, the descriptor mediates by storing reference information that allows multiple queues to access the same cache node, thereby reducing the total number of cache nodes needed while maintaining the ability to distribute messages to multiple destinations.
3Ease of operation
If two sets of linked lists are used to manage packet entities and descriptors, then both packet and queue management can be performed, but the maintenance overhead and processing complexity increase
Solution Approach 1:
The patent combines the two separate linked list management structures (one for packet entities and one for descriptors) into a single unified linked list. Each node in this unified structure contains both the packet entity information and the descriptor information, allowing both types of management operations to be performed through a single data structure, thereby reducing maintenance overhead and processing complexity.
Data Source
Figure 1~3
Figure 4~5
Figure 6
AI summary
A message processing method and device, the method comprises: allocating a node to an input message in a cache space for saving the message, and taking the location corresponding to the cache space as index information of a descriptor of the message; and extracting the descriptor information of the message; framing and saving the descriptor information and node information of the message in a node linked list. The abovementioned solution can achieve uniform memory of unicast and multicast messages, and the descriptor linked list corresponds to the packet entity caching resource, thereby significantly reducing the overheads for managing the unicast and multicast messages, so as to improve the node aggregation capability.