Network Processor Exception Management via Signature Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network processors face excessive strain and dropped packets due to frequent exceptions of similar packets, leading to retransmissions and increased network traffic when the initial packet's result is not found quickly, causing other hosts' data to be dropped.
Innovation Solution
Implementing a method and apparatus that uses a signature table and secondary table within the network processor to manage exceptions by hashing packet information, queuing packets with the same signature, and limiting the number of outstanding searches to prevent overwhelming the core processor, thereby reducing retransmissions and network strain.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the core processor processes every exception packet individually, then each packet receives proper handling, but the core processor becomes overwhelmed and network throughput decreases
Solution Approach 1:
The patent merges multiple exception packets with the same signature into a single aggregated exception record. Instead of processing each packet individually, the core processor receives one combined exception that represents multiple original packets, thereby reducing processing overhead while maintaining complete exception handling.
Solution Approach 2:
The patent creates a signature-based copy representation of exception packets. Rather than transferring and processing the actual packet data multiple times, the system uses compact signature copies to identify and group similar exceptions, reducing the data volume that the core processor must handle.
2Speed
If the cached SPD database is used for quick lookup, then processing speed increases, but packets with cache misses require full SPD database searches that cause delays
Solution Approach 1:
The patent performs preliminary aggregation of exception packets in the processing element before they reach the core processor. By grouping packets with the same signature in advance and creating a single aggregated exception record, the system prepares the data structure that will minimize core processor search time and prevent repeated full SPD database searches.
3Ease of operation
If similar exception packets are processed separately, then each packet is handled individually, but network strain increases due to retransmissions of dropped packets
Solution Approach 1:
The patent combines multiple similar exception packets into a single aggregated exception record that preserves all necessary information for proper handling. This merging prevents packet drops caused by core processor overload while maintaining the simplicity of individual packet processing, as the aggregation logic automatically manages the grouping and the core processor handles the aggregated exception as a unified entity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively manages exceptions by queuing packets with the same signature, reducing the need for retransmissions and minimizing network strain, thereby maintaining network throughput and efficiency.
Implementation Method 1
A hash unit 40 is used by the core processor 30 and the processing elements 20 to perform hash calculations
Data Source
AI summary
A method and apparatus to process exception packets is described. A signature table, secondary table and a plurality of queues are used to store exception packets which match packets that are currently being handled, such that multiple requests for exception packet handling of the same or similar packets do not overwhelm the network processor.