SD-WAN Router PMTU-Aware Flow Rerouting for Fragmentation Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data packet fragmentation in SD-WAN routers leads to processing overhead and performance issues due to fluctuating Path Maximum Transfer Units (PMTU) across egress links, causing excessive fragmentation and reassembly, which induces jitter and processing load.
Innovation Solution
Implementing Data Fragmentation Logic in SD-WAN routers to monitor and manage packet fragmentation rates on a per-flow and per-link basis, rerouting data flows to egress links with higher PMTU to prevent excessive fragmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data packets are fragmented to fit egress link PMTU, then data transmission is enabled, but processing overhead and jitter increase
Solution Approach 1:
The system performs preliminary actions by detecting PMTU values for all egress links in advance and storing them in a database. Before data flows are routed, the system queries the appropriate PMTU for each potential egress link, allowing routing decisions to be made with full knowledge of capacity constraints, thereby avoiding fragmentation later in the transmission process
Solution Approach 2:
The system dynamically adapts routing decisions based on real-time PMTU conditions. The AAR algorithm is enhanced to query current PMTU values and adjust flow-to-link mappings dynamically. When PMTU conditions change or excessive fragmentation is detected, the system re-routes flows to optimal egress links, making the routing flexible and adaptive rather than static
2Productivity
If Application Aware Routing assigns data flows to egress links, then routing efficiency is improved, but fragmentation rate increases when PMTU is exceeded
Solution Approach 1:
The system implements feedback mechanisms where fragmentation statistics are continuously monitored and fed back to the routing algorithm. When excessive fragmentation is detected on a particular egress link, this information is used to adjust future routing decisions, de-prioritizing or re-routing flows that cause fragmentation. The system also queries PMTU values dynamically to inform routing decisions, creating a closed-loop control system
Solution Approach 2:
The system changes routing parameters by modifying which egress link is selected for each data flow based on PMTU conditions. Instead of using fixed routing rules, the AAR algorithm queries PMTU values and selects egress links whose PMTU can accommodate the data flow size. This parameter-based routing approach ensures that flows are assigned to links with sufficient capacity, preventing fragmentation
3Adaptability or versatility
If multiple egress links with different PMTU values are used, then transport independence is achieved, but fragmentation management complexity increases
Solution Approach 1:
The system achieves universality by creating a unified fragmentation management approach that works across all egress links regardless of their different PMTU values. The enhanced AAR algorithm serves multiple functions: it performs application-aware routing, queries PMTU values, selects appropriate egress links, and monitors fragmentation statistics. This multi-functional approach simplifies complexity by providing a single coherent mechanism rather than separate handling for each link type
Data Source
AI summary
Techniques are described for avoiding data packet fragmentation in a routing device such as a router or network switch. Path Maximum Transport Unit (PMTU) values are monitored for a plurality of egress links of a networking device. A statistical analysis of fragmentation rates is performed. The statistical analysis can be performed on a per-link basis, per-flow basis or both per-link and per-flow basis. If the packet fragmentation rate of data flows through a particular egress link exceeds a determined threshold value, one or more data flows can be re-routed to a different egress link having a higher PMTU, thereby preventing data packet fragmentation.