MAC Layer Packet Fragmentation for Ultra-Low Latency Forwarding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional packet forwarding devices struggle to meet the requirement of ultra-low latency for latency-sensitive services due to high forwarding latency, especially in traffic convergence scenarios where the forwarding of a current packet is delayed until a previous packet is fully sent, and existing data transmission technologies fail to reduce latency at the MAC layer for packet forwarding.
Innovation Solution
The method involves buffering fragmented data frames into separate queues based on their latency sensitivity at the MAC layer, with latency-sensitive frames being processed prior to latency-insensitive frames, and using a combination of fragmenting and reassembly techniques to reduce forwarding latency, including the use of latency-sensitive and latency-insensitive labels to prioritize and manage data frames effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If store-and-forward manner is used, then forwarding accuracy is ensured, but forwarding latency is long
Solution Approach 1:
The patent segments data packets into smaller fragments at the MAC layer, allowing individual fragments to be forwarded independently rather than waiting for complete packets. This segmentation enables cut-through switching of fragments while maintaining the ability to reassemble at the destination, thus reducing forwarding latency while ensuring accuracy through proper reassembly.
Solution Approach 2:
The patent performs preliminary fragmenting of data packets before forwarding, creating fragment headers with forwarding information in advance. This allows the forwarding device to immediately forward fragments without waiting for complete packet reception, achieving preliminary action that reduces latency while maintaining forwarding accuracy through pre-prepared fragment structures.
2Loss of time
If cut-through switching is used, then forwarding latency is reduced, but forwarding of subsequent packets is delayed in traffic convergence scenarios
Solution Approach 1:
By segmenting packets into fragments that can be forwarded independently, the patent allows multiple fragments from different original packets to be interleaved in the forwarding queue. This segmentation prevents a single long packet from blocking subsequent packets, as each fragment is a smaller, independently forwardable unit, thus improving throughput while maintaining low latency.
Solution Approach 2:
The patent introduces dynamic fragment-level scheduling that can prioritize and interleave fragments from different packets based on queue status and latency requirements. This dynamic approach allows the system to adaptively manage traffic convergence scenarios, preventing blocking and improving overall packet forwarding throughput while maintaining low latency for time-sensitive fragments.
3Adaptability or versatility
If network layer fragmenting is used, then data transmission flexibility is improved, but MAC layer forwarding latency cannot be reduced
Solution Approach 1:
The patent performs segmentation at the MAC layer rather than the network layer, creating MAC layer fragments with MAC layer headers containing forwarding information. This MAC layer segmentation enables the forwarding device to process and forward fragments at the MAC layer without needing to wait for network layer processing, thus reducing MAC layer forwarding latency while maintaining transmission flexibility through proper fragment management.
Solution Approach 2:
The patent introduces MAC layer fragment headers as an intermediary structure that carries necessary forwarding information at the MAC layer level. This intermediary enables forwarding decisions to be made at the MAC layer without requiring network layer involvement, thus reducing MAC layer latency while maintaining the flexibility of data transmission through proper fragment reassembly at the destination.
Data Source
Figure 1~2
Figure 3A
Figure 3B
AI summary
This application discloses a packet processing method and apparatus, to resolve a problem that an ultra-low latency forwarding requirement cannot be met because a forwarding latency of a latency-sensitive service is relatively long. The method includes: obtaining, by a first network device at a MAC layer, a first fragmented data frame included in a first data frame, and buffering the first fragmented data frame into a first queue; obtaining, by the first network device at the MAC layer, a second fragmented data frame included in a second data frame, and buffering the second fragmented data frame into a second queue; sending, by the first network device, the first fragmented data frame to a forwarding processing module, and obtaining first forwarding information by using the forwarding processing module; and after the first network device sends the first fragmented data frame to the forwarding processing module, sending the second fragmented data frame to the forwarding processing module, and obtaining second forwarding information by using the forwarding processing module. Therefore, in the method provided in this application, a forwarding latency of a latency-sensitive service in a network device can be reduced, and it can be ensured that the latency-sensitive service is not affected by a latency-insensitive service.