L4S Queuing and Marking for Low Latency Networking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computer networking technologies face challenges in integrating Low Latency, Low Loss, Scalable Throughput (L4S) queuing and marking with legacy devices, particularly in supporting L4S requirements such as scalable congestion control, explicit congestion notification, and conditional priority scheduling.
Innovation Solution
The implementation of L4S queuing and marking involves determining congestion in layer 2, sending a marking request to layer 3, marking packets for congestion, and returning the marked packets to layer 2. Additionally, a new congestion flag is introduced to indicate predicted congestion events, allowing for proactive congestion avoidance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional queuing mechanisms are used in legacy network devices, then device compatibility is maintained, but queuing latency increases and throughput scalability is limited
Solution Approach 1:
The patent segments the queue into multiple priority levels (e.g., high priority and low priority queues) within the network device. This segmentation allows L4S traffic to be separated from traditional traffic, enabling low-latency handling for L4S packets while maintaining backward compatibility with legacy devices that handle standard traffic in the traditional queue.
Solution Approach 2:
The patent implements dynamic queue management where the network device can adaptively adjust queue parameters based on traffic conditions and L4S packet markings. This includes dynamic priority assignment and queue size adjustment, allowing the system to optimize for low latency when L4S traffic is detected while maintaining stable operation for legacy traffic.
2Reliability
If congestion marking is implemented at layer 3 only, then implementation simplicity is maintained, but congestion detection accuracy and proactive avoidance capability are reduced
Solution Approach 1:
The patent implements congestion detection and marking at layer 2 (data link layer) before packets reach layer 3 (network layer). This preliminary action allows the network device to identify congestion conditions earlier in the packet processing pipeline, mark packets proactively before they are forwarded, and prevent congestion from propagating further through the network.
Solution Approach 2:
The patent introduces an intermediary mechanism between layer 2 and layer 3 that coordinates congestion marking operations. This intermediary layer translates congestion detection from layer 2 into appropriate marking actions, enabling accurate congestion detection while managing the complexity of multi-layer coordination through a standardized interface.
3Reliability
If reactive congestion control is used, then system simplicity is maintained, but latency and loss performance deteriorate under congestion conditions
Solution Approach 1:
The patent implements proactive congestion marking where network devices predict potential congestion conditions and mark packets before actual congestion occurs. This preliminary marking allows end hosts to take preventive actions (such as reducing transmission rate) before queue overflow and packet loss occur, achieving low loss performance without requiring complex reactive control mechanisms.
Solution Approach 2:
The patent establishes a feedback loop where congestion markings on forwarded packets are communicated back to the transmitting host. This feedback mechanism enables the host to adjust its transmission behavior in real-time based on network conditions, achieving reliable low-loss performance through coordinated control while keeping individual device complexity manageable.
Data Source
AI summary
Low Latency, Low Loss, Scalable Throughput (L4S) queuing and marking and, specifically, L4S parameters for congestion marking and a new congestion flag for predicted congestion events may be provided. Providing L4S marking can comprise determining, in layer 2, a packet is experiencing congestion. A L4S marking request is sent to layer 3 to request congestion marking of the packet. The packet is marked in layer 3 for congestion experienced A L4S marking response comprising the marked packet is then sent to layer 2.