RSS Key Selection Using Weighted Flow Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current receive side scaling (RSS) methods face inefficiencies in distributing network packets across multiple processors, leading to latency spikes and reduced performance due to limited key generation and reliance on Network Layer 3 and Transport Layer 4 information, causing network packets to be stuck in full queues.
Innovation Solution
A method is introduced to optimize RSS key selection by assigning weights to different fields of flow data, generating scores for balanced processor loading, and updating the current RSS key with an optimal key based on aggregated network connection flow data, excluding low-impact flows to conserve resources and improve packet distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional RSS key generation methods are used based on Network Layer 3 and Transport Layer 4 information, then the implementation is simple, but the packet distribution is unbalanced causing latency spikes and reduced performance
Solution Approach 1:
The patent segments the flow data into multiple fields (source IP, destination IP, source port, destination port, protocol) and assigns different weights to each field. This segmentation allows the system to evaluate multiple dimensions of packet flow characteristics separately, enabling more balanced distribution across processors while maintaining a structured and manageable key generation process.
Solution Approach 2:
The patent changes the parameters used for RSS key generation from traditional L3/L4 information to a weighted combination of multiple flow data fields. By introducing weight parameters that can be adjusted to reflect the importance of different flow characteristics, the system achieves better load balancing while keeping the underlying hash function mechanics unchanged.
2Measurement precision
If all network connection flows are considered for RSS key optimization, then the distribution accuracy improves, but the computational resources are wasted on low-impact flows
Solution Approach 1:
The patent applies partial action by selectively considering only high-impact network flows for RSS key optimization. A threshold mechanism filters out low-impact flows, allowing the system to focus computational resources on the subset of flows that actually affect load balancing performance, thereby avoiding waste on insignificant traffic.
Solution Approach 2:
The patent applies local quality by treating different network flows differently based on their impact characteristics. High-impact flows receive detailed analysis and weighted field evaluation, while low-impact flows are filtered out. This localized attention to quality ensures accurate distribution for important traffic without the overhead of processing all flows uniformly.
3Stability of the object's composition
If weighted field evaluation is applied to generate RSS key scores, then the processor loading balance improves, but the calculation complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating weights for different flow data fields based on their importance to load balancing. These weights are determined in advance and stored, so during runtime the system only needs to retrieve and apply them rather than performing complex optimization calculations for each packet, thus achieving stable processor loading with minimal real-time computational overhead.
Data Source
AI summary
Optimizing receive side scaling (RSS) key selection is provided. Different weights are assigned to different fields of flow data corresponding to a network connection of a registered client device. A score is generated representing an amount of balanced processor loading for each RSS key corresponding to the registered client device based on the different fields of the flow data with assigned weights. A current RSS key on the registered client device is updated with an optimal RSS key based on the score corresponding to the optimal RSS key representing balanced loading of processors on the registered client device.