Multi-Stage Switch Packet Classification Using Compressed Policy Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing packet classification schemes, such as those based on TCAMs or software algorithms, face challenges with power consumption, chip area requirements, and memory bandwidth limitations, making them impractical for high-speed switches and routers, particularly in complex routing systems like data centers.
Innovation Solution
A packet classification module at a multi-stage switch uses a policy vector module to retrieve and decompress a compressed policy vector from a database, defining a decompressed policy vector that triggers processing based on bit values associated with policy conditions, allowing efficient classification and routing of data packets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If known packet classification schemes (TCAMs or software algorithms) are used, then packet classification functionality is achieved, but power consumption and chip area requirements become undesirable
Solution Approach 1:
The patent divides the packet classification task into multiple stages, with each stage handling a portion of the classification process. This segmentation allows the system to achieve comprehensive packet classification functionality while using simpler, more power-efficient components at each stage, avoiding the need for large-power-consuming TCAMs or complex software algorithms in a single stage.
Solution Approach 2:
The patent transitions from traditional single-stage packet classification approaches to a multi-stage architecture, adding a temporal dimension to the classification process. This dimensional change enables the system to distribute classification logic across multiple stages, reducing the power and area requirements of individual stages while maintaining overall classification effectiveness.
2Productivity
If software packet classification algorithms with large data structures in external memory are used, then comprehensive packet classification is achieved, but memory bandwidth limitations prevent use in very high speed switches and routers
Solution Approach 1:
The patent performs preliminary actions by pre-processing and organizing classification data structures in a manner that minimizes subsequent memory access requirements. The multi-stage architecture prepares classification information in advance at each stage, reducing the need for high-speed external memory access and enabling the system to operate at very high speeds without being bottlenecked by memory bandwidth.
3Adaptability or versatility
If packet classification schemes with high power consumption and inefficient chip designs are used, then packet classification functionality is achieved, but scaling for use in complex routing systems such as data centers is prevented
Solution Approach 1:
The patent implements a dynamic multi-stage packet classification architecture where each stage can be independently configured and scaled. This dynamic design allows the system to adapt to different routing complexity requirements by adjusting the number and configuration of stages, while maintaining power efficiency at each stage. The modular nature enables incremental scaling for complex data center routing systems without requiring a complete redesign or incurring prohibitive power costs.
Data Source
AI summary
In one embodiment, an apparatus can include a policy vector module configured to retrieve a compressed policy vector based on a portion of a data packet received at a multi-stage switch. The apparatus can also include a decompression module configured to receive the compressed policy vector and configured to define a decompressed policy vector based on the compressed policy vector. The decompressed policy vector can define a combination of bit values associated with a policy.