Multi-Column Traffic Statistics Memory for Packet Flow Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In packet switched networks, existing methods for testing and analyzing network traffic face inefficiencies due to collisions when using hash functions to map flow identifiers to address pointers, leading to incorrect accumulation of traffic statistics and reduced ability to track all flows, especially as the number of unique flow identifiers approaches the number of available records in the statistics memory.
Innovation Solution
The implementation of a traffic receiver that uses a combination of extracted fields from packets to generate a flow identifier, applying masks and a hash function to access traffic statistics memory organized in multiple columns, reducing collisions and improving the ability to accumulate statistics across all flows by optimizing the data path width and record organization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a hash function is used to map flow identifiers to address pointers in traffic statistics memory, then the ability to accumulate statistics for multiple flows is improved, but collisions occur when the number of flow identifiers approaches the number of available records, leading to incorrect statistics accumulation
Solution Approach 1:
The traffic statistics memory is divided into multiple columns, with each column containing multiple records. This segmentation allows the system to track a larger number of flows by distributing flow identifiers across different columns, reducing collisions. When a flow identifier is hashed, it can map to multiple potential records across different columns, and the system checks each column to find the correct record or determine if it's a new flow.
Solution Approach 2:
The patent introduces an additional dimension to the memory structure by organizing records in multiple columns rather than a single linear array. This multi-dimensional organization increases the effective capacity for tracking flows without requiring a proportional increase in the total number of records, as the same records can serve multiple flow identifiers from different columns.
2Adaptability or versatility
If the number of flow identifiers increases to cover more network traffic patterns, then the comprehensiveness of traffic analysis is improved, but the collision rate in hash mapping increases, reducing tracking accuracy
Solution Approach 1:
By segmenting the memory into multiple columns, the system can accommodate a larger number of flow identifiers while maintaining measurement precision. Each column acts as an independent tracking space, allowing the system to monitor more flows simultaneously without increasing the collision rate within any single column.
Solution Approach 2:
The patent uses column identifiers as an intermediary layer between the hash function output and the actual record access. Instead of directly mapping a flow identifier to a single record, the system first determines which column to access based on the flow identifier, then applies the hash function within that column context, reducing direct collisions.
3Productivity
If traditional single-column memory structure is used, then the device complexity is low, but the ability to accumulate statistics for all flows is limited when memory capacity is approached
Solution Approach 1:
The memory structure is segmented into multiple columns, each capable of independently storing flow statistics records. This segmentation increases the overall flow tracking capacity without requiring a complete redesign of the memory architecture, as each column can be managed independently while working together to provide comprehensive flow coverage.
Solution Approach 2:
The multi-column memory structure serves multiple functions: it maintains the simplicity of single-column access for individual flows while providing the capacity of a much larger single-column structure. Each column can be accessed using the same basic operations, making the system universal in its approach while scaling to handle more flows.
Data Source
AI summary
Methods, apparatus, and computer readable storage media for accumulating traffic statistics. A traffic receiver may receive a packet and determine a flow identifier associated with the received packet. A hash function may be applied to the flow identifier to generate a memory pointer. One or more records may be retrieved from a statistics memory based on the memory pointer. A determination may be made if the flow identifier associated with the received packet matches a flow identifier value contained in one of the one or more retrieved records. When a match occurs, the retrieved record containing the matching flow identifier value may be updated and the updated record may be stored in the statistics memory.