Packet Memory Segmentation for Network Device Power Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional network devices store entire packets in memory for processing and transmission, leading to inefficient use of memory and increased power consumption due to the need for storing information for header modification in transmit queues.

Innovation Solution

A method where network devices store a first portion of the packet, including the payload, and a second portion, including a portion of the header, separately in memory, allowing for efficient processing and transmission by combining these portions when needed, and using a 'lean' packet descriptor for enqueuing that omits information for header modification, reducing memory and power requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the entire packet is stored in packet memory for processing and transmission, then the packet can be fully processed and forwarded, but memory usage increases and power consumption rises due to storing header modification information in transmit queues

Engineering Contradiction:
Improvepacket processing completenessVSAvoidmemory usage
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments the packet into two separate portions: the payload portion is stored in packet memory by the receive processor, while the header portion is processed separately by the packet processor. This segmentation eliminates the need to store complete packets in memory, reducing memory usage while maintaining processing reliability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts the header portion from the complete packet and processes it separately through the packet processor. By taking out the header for separate processing and using a lean packet descriptor that omits header modification information, the system reduces the quantity of data that needs to be stored in transmit queues, thereby reducing memory usage and power consumption.

Inventive Principle:
Principle #2Taking out (Extraction)

2Ease of operation

If header alteration is performed after packet retrieval from memory, then transmit queues need to store information for header modification, but this increases power consumption and memory requirements

Engineering Contradiction:
Improveheader modification flexibilityVSAvoidpower consumption
Core Design Contradiction:
Ease of operationVSUse of energy by stationary object

Solution Approach 1:

The patent employs a lean packet descriptor that is intentionally minimal and disposable, containing only essential transmission information while omitting header modification information. This lightweight descriptor reduces the amount of data that needs to be stored and processed in transmit queues, thereby reducing power consumption while maintaining the flexibility to perform header alterations when needed.

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

3Manufacturing precision

If complete packet information is stored in transmit queues for header modification, then header alteration can be performed accurately, but the communication bus width increases

Engineering Contradiction:
Improveheader modification accuracyVSAvoidcommunication bus width
Core Design Contradiction:
Manufacturing precisionVSLength of moving object

Solution Approach 1:

The patent extracts only the essential transmission information from the complete packet and places it in the lean packet descriptor. By taking out only the necessary fields (such as destination address and transmission control information) and omitting header modification information, the system reduces the data width that must be transmitted over the communication bus, thereby narrowing the bus width while maintaining sufficient information for accurate header alteration when required.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentUS10764410B2Method and apparatus for processing packets in a network device
Publication Date: 2020.09.01 MARVELL ISRAEL (M L S L) LTD
  • US10764410B2 patent drawing
  • US10764410B2 patent drawing
  • US10764410B2 patent drawing

AI summary

A packet received by a network device via a network. A first portion of the packet is stored in a packet memory, the first portion including at least a payload of the packet. The packet is processed based on information from a header of the packet. After the packet is processed, a second portion of the packet is stored in the packet memory, the second portion including at least a portion of the header of the packet. When the packet is to be transmitted the first portion of the packet and the second portion of the packet are retrieved from the packet memory, and the first portion and the second portion are combined to generate a transmit packet. The transmit packet is forwarded to a port of the network device for transmission of the transmit packet via port of the network device.