Network Interface Controller Dynamic Packet Flow Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Increased network traffic and connection speeds put a significant burden on host processor resources due to frequent memory accesses for TCP control block data, which can degrade system performance, especially in multi-processor systems where maintaining cache consistency is challenging.
Innovation Solution
Implementing a network interface controller that learns to direct packets by identifying sources of egress packets and updating its flow mapping, allowing it to route ingress packets efficiently without explicit configuration messages, thereby reducing runtime configuration and adapting quickly to changes in processor and queue assignments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If TCP operations are handled by host processors, then connection management and data transfer reliability are maintained, but processor resources are heavily burdened by frequent memory accesses for TCP control block data
Solution Approach 1:
The patent extracts TCP control block data from main memory and places it in processor-specific cache memory. This extraction reduces the burden on host processors by minimizing frequent memory accesses, thereby improving processor performance while maintaining connection management reliability.
Solution Approach 2:
The system performs preliminary actions by pre-loading TCP control block data into cache memory before it is needed for packet processing. This preliminary caching action reduces latency during packet handling and improves overall processor productivity without compromising connection management.
2Productivity
If TCP control block data is cached in processor-specific cache memory, then memory access frequency is reduced and processor performance improves, but cache consistency becomes difficult to maintain in multi-processor systems
Solution Approach 1:
The patent segments TCP control block data by associating each processor with its own dedicated cache memory for storing TCP control blocks. This segmentation allows each processor to independently manage its cache without consistency conflicts, maintaining both processor performance and cache stability in multi-processor systems.
Solution Approach 2:
Each processor has local access to its own cached TCP control block data, providing local quality optimization. This local caching strategy eliminates the need for inter-processor cache synchronization, thereby maintaining cache consistency while improving processor performance.
3Ease of operation
If explicit configuration messages are used to direct packet flows to processors, then packet routing is controlled, but runtime configuration overhead and bus traffic increase
Solution Approach 1:
The network interface controller performs self-service by autonomously learning and adapting packet flow routing based on observed packet patterns. This eliminates the need for explicit configuration messages, reducing runtime configuration overhead and bus traffic while maintaining ease of packet routing control.
Solution Approach 2:
The system implements feedback mechanisms where the network interface controller monitors packet flows and automatically adjusts routing decisions based on observed patterns. This feedback-driven adaptive routing eliminates configuration messages and reduces setup time while maintaining routing control.
4Adaptability or versatility
If packet flows are dynamically assigned to processors based on learned patterns, then adaptation to processor and queue assignments is immediate, but the complexity of flow mapping management increases
Solution Approach 1:
The patent implements dynamic flow mapping where the network interface controller continuously learns and adapts packet flow routing based on observed patterns and system state. This dynamic adaptation enables immediate response to processor and queue assignments while the learning algorithms manage the complexity of flow mapping automatically.
Data Source
AI summary
In general, in one aspect, the disclosure describes a method includes accessing data of an egress packet belonging to a flow, storing data associating the flow with at least one queue based on a source of the data of the egress packet. The method also includes accessing an ingress packet belonging to the flow, performing a lookup of the at least one queue associated with the flow, and enqueueing data of the ingress packet to the at least one queue associated with the flow.