UDP Segmentation Offload via IP Header Sequencing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
UDP traffic segmentation offload is challenging due to the lack of inherent sequencing in UDP payloads, leading to inaccurate reconstruction of original payloads at the receiver, especially in high bandwidth networks, where CPU utilization is high and conventional solutions require application awareness and rewriting.
Innovation Solution
Modifying legacy header fields in Ethernet and IP headers to encode sequencing information, allowing the network stack to process sequenced frames transparently, reducing CPU utilization without requiring application updates, and enabling seamless communication across configured and unconfigured receivers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If segmentation offload is implemented for UDP traffic, then CPU utilization is reduced, but accurate reconstruction of original payloads becomes impossible due to lack of sequencing information
Solution Approach 1:
The patent divides a large UDP payload into multiple MTU-sized segments for transmission, with each segment being processed independently by network hardware. This segmentation enables offload processing while maintaining the ability to reconstruct the original payload through sequence numbers embedded in each segment's IP header.
Solution Approach 2:
The patent embeds sequence numbers in the IP header fields before segmentation occurs. This preliminary encoding of sequencing information allows the receiver to correctly reassemble segments even when they arrive out of order or are lost, without requiring application-layer intervention.
2Reliability
If metadata is encoded in the payload to enable out-of-order detection, then payload reconstruction accuracy improves, but application rewriting is required and processing complexity increases
Solution Approach 1:
The patent uses the IP header as an intermediary to carry sequencing information. Instead of requiring applications to embed metadata in the payload, the network stack automatically inserts sequence numbers into IP header fields, transparently enabling reliable reconstruction without application modifications.
Solution Approach 2:
The network stack automatically performs segmentation, sequence number insertion, and reassembly operations without requiring application-layer awareness or intervention. This self-service approach maintains reliability while avoiding the complexity of application rewriting.
3Ease of operation
If legacy header fields are modified to encode sequencing information, then transparency to applications is maintained and CPU utilization is reduced, but compatibility with unconfigured receivers may be affected
Solution Approach 1:
The patent reuses existing IP header fields (such as identification and fragment offset fields) for dual purposes: both their original functions and for carrying sequencing information. This multi-functionality maintains compatibility with standard receivers while enabling segmentation offload with sequencing for configured receivers.
Data Source
AI summary
A user datagram protocol (UDP) is a well-known protocol for transferring data between two nodes of a network. When data is too large to fit within a single UDP packet that can be transmitted between the two nodes, the data needs to be segmented and transmitted with multiple packets and reassembled on the receiving node. Techniques are disclosed herein, for example, for offloading such segmentation, transmission, and reassembly from the central processing units (CPUs) of the nodes. Such offloading is performed efficiently, for example, by repurposing legacy protocol fields used in UDP transmission such as the internet protocol (IP) identification (ID), time to live (TTL), type of service (TOS), and/or EtherType fields to encode information needed for efficient segmentation, out of order reception, and reassembly.