Adaptive Drop-Tail Active Queue Management for Network Switches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional Droptail queue management in network equipment leads to long queuing delays and unnecessary packet loss due to delayed congestion notification, resulting in reduced link utilization and increased latency, as it only drops packets when buffers overflow, causing senders to react aggressively and reduce transmission rates.
Innovation Solution
Adaptive Drop-Tail, a queue-based active queue management process that applies congestion notification by adjusting the queue size threshold based on congestion levels, increasing buffer capacity during increased congestion and decreasing it during decreased congestion, thereby reducing unnecessary packet drops and maintaining high link utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If Droptail queue management is used, then implementation is simple, but packet loss rate increases and queuing delay increases
Solution Approach 1:
The patent applies preliminary action by dropping packets before the queue reaches maximum capacity. When queue length exceeds a threshold (e.g., 80% of buffer size), packets are proactively dropped with probability proportional to queue length, preventing buffer overflow and reducing packet loss rates before they occur.
Solution Approach 2:
The patent implements feedback by continuously monitoring queue length and adjusting packet drop probability dynamically. The drop probability increases as queue length increases, creating a feedback mechanism that regulates traffic flow and prevents congestion, thereby reducing both packet loss and queuing delay.
2Device complexity
If Droptail queue management is used, then buffer overflow handling is simple, but link utilization decreases
Solution Approach 1:
The patent applies preliminary action by implementing early congestion notification through proactive packet dropping before buffer overflow occurs. This prevents the aggressive rate reduction by senders that happens with Droptail, maintaining higher link utilization by avoiding unnecessary transmission rate cuts.
Solution Approach 2:
The patent implements feedback by providing continuous congestion signals to senders through packet drops proportional to queue length. This gradual feedback allows senders to reduce rates smoothly rather than aggressively, maintaining better link utilization while still managing congestion effectively.
3Device complexity
If Droptail queue management is used, then congestion notification is delayed, but implementation complexity is low
Solution Approach 1:
The patent applies preliminary action by notifying senders of congestion conditions before the queue reaches maximum capacity. Packets are dropped when queue length exceeds a threshold (e.g., 80% of buffer), providing early congestion notification that allows senders to react before severe congestion occurs, reducing notification delay.
Solution Approach 2:
The patent implements feedback by continuously monitoring queue length and providing real-time congestion signals through packet drops. This continuous feedback mechanism reduces notification delay by immediately signaling congestion conditions to senders, allowing faster reaction times compared to Droptail's delayed overflow-based notification.
4Productivity
If queue size threshold is increased to accommodate more packets, then throughput increases, but queuing delay increases
Solution Approach 1:
The patent applies dynamics by making the effective queue size dynamic through probabilistic packet dropping. While the physical buffer size remains fixed, the effective queue size varies based on current load conditions, allowing the system to adapt between maximizing throughput and minimizing delay based on real-time congestion levels.
Solution Approach 2:
The patent changes parameters by dynamically adjusting packet drop probability based on queue length. This parameter change allows the system to effectively reduce queue size during congestion (lowering delay) while maintaining larger effective buffer capacity during light load (maximizing throughput), resolving the throughput-delay tradeoff.
Data Source
AI summary
An active queue management (AQM) process for network communications equipment. The AQM process is queue based and involves applying at a queue size threshold congestion notification to communications packets in a queue of a link via packet dropping; and adjusting said queue size threshold on the basis of the congestion level. The AQM process releases more buffer capacity to accommodate more incoming packets by increasing said queue size threshold when congestion increases; and decreases buffer capacity by reducing said queue size threshold when congestion decreases. Network communications equipment includes a switch component for switching communications packets between input ports and output ports, packet queues for at least the output ports, and an active queue manager for applying congestion notification to communications packets in the queues for the output ports via packet dropping. The congestion notification is applied at respective queue size thresholds for the queues, and the thresholds adjusted on the basis of the respective congestion levels of the queues of the output ports.