NTU Queuing System Buffer Segmentation Bandwidth Optimization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Bulk storage devices like SDRAM can only be accessed efficiently in large fixed-size chunks, leading to inefficient storage and bandwidth utilization when handling frames or packets of varying sizes, particularly in network termination equipment, where up to half the available bandwidth is wasted.

Innovation Solution

A queuing system that splits the bulk storage device into buffers, allowing frames to be packed into fixed-size buffers and stored only when complete, with a link label mechanism to link segments and manage bandwidth efficiently, enabling multiple flows to write to any queue and supporting time division multiplexed data streams.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If frames of varying sizes are stored directly in bulk storage device, then storage flexibility is improved, but bandwidth utilization deteriorates due to inefficient access patterns

Engineering Contradiction:
Improvestorage flexibilityVSAvoidbandwidth utilization
Core Design Contradiction:
Adaptability or versatilityVSLoss of energy

Solution Approach 1:

The bulk storage device is segmented into multiple fixed-size buffers, each capable of holding a complete frame or a segment of a frame. This segmentation allows the system to maintain fixed-size access patterns for efficient bandwidth utilization while still accommodating variable-size frames through the buffer structure and segment linking mechanism.

Inventive Principle:
Principle #1Segmentation

2Loss of energy

If frames are packed into fixed-size buffers, then bandwidth utilization is improved, but frame completeness control deteriorates

Engineering Contradiction:
Improvebandwidth utilizationVSAvoidframe completeness control
Core Design Contradiction:
Loss of energyVSReliability

Solution Approach 1:

The system employs feedback mechanisms through link labels attached to each buffer. These link labels contain information about frame segments and their relationships, enabling the system to track and verify frame completeness. When a frame is reassembled from segments, the feedback from link labels ensures that only complete frames are processed, maintaining reliability while achieving efficient bandwidth utilization through fixed-size buffer packing.

Inventive Principle:
Principle #23Feedback

3Adaptability or versatility

If multiple flows are allowed to write to any queue, then system versatility is improved, but queue management complexity deteriorates

Engineering Contradiction:
Improvesystem versatilityVSAvoidqueue management complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The queue structure is designed with universal characteristics where any input flow can write to any queue, and any queue can serve multiple output flows. This multi-functionality is achieved through the standardized buffer structure with link labels that can represent any frame segment from any flow. The system manages this versatility through a unified buffer management mechanism rather than dedicated queues for each flow, reducing overall complexity while maintaining flexibility.

Inventive Principle:
Principle #6Universality (Multi-functionality)

4Adaptability or versatility

If segments are stored in linked lists, then frame reassembly flexibility is improved, but memory overhead deteriorates

Engineering Contradiction:
Improveframe reassembly flexibilityVSAvoidmemory overhead
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The system merges the frame data storage with the link label information in a unified buffer structure. Instead of separate structures for frame data and linking information, each buffer contains both the frame segment data and the link label that provides reassembly information. This merging eliminates redundant memory structures and reduces overhead while maintaining the flexibility of linked list-based frame reassembly through the integrated link labels.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP2015521B1NTU queuing system
Publication Date: 2011.01.26 ADVA OPTICAL NETWORKING SP ZOO
  • EP2015521B1 patent drawingFigure 1~2
  • EP2015521B1 patent drawingFigure 3~4
  • EP2015521B1 patent drawingFigure 5~6

AI summary

The invention relates to a method for storing frames or other packets of data in queues in a bulk storage device comprising the steps of: splitting up at least a portion of the bulk storage device into a number of buffers each of one or more chunks and queuing the frames or other packets of data which arrive from an input pipe as a series of segments, wherein a frame or other packet of data may be split into one or more segments, wherein if a frame or other packet of data is split into multiple segments, each segment except the last will fill a buffer.