Per-Flow Network Queues for Fast Endpoint Congestion Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network architectures face challenges in scalability, versatility, and efficiency due to increasing network load and diverse traffic types, with conventional congestion control mechanisms being slow and ineffective, especially in large or heavily loaded networks.
Innovation Solution
A data-driven intelligent networking system with per-flow congestion control that maintains state information of individual packet streams, using flow-specific input queues and acknowledgments to enable dynamic setup and teardown of flow channels, allowing switches to perform endpoint congestion detection and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional congestion control mechanisms are used, then network architecture is simpler, but congestion control speed and effectiveness deteriorate in large or heavily loaded networks
Solution Approach 1:
The patent segments the network into per-flow queues at each switch, maintaining separate state information for each flow rather than using aggregate network-wide congestion control. This segmentation enables faster, flow-specific congestion detection and control while distributing the complexity across multiple switches rather than requiring a centralized complex controller.
Solution Approach 2:
The patent implements dynamic flow setup and teardown based on injected data, where flow state information is maintained dynamically at each switch. This dynamic approach allows the network to adapt to changing traffic patterns in real-time, improving congestion control speed while the distributed nature keeps individual switch complexity manageable.
2Measurement precision
If per-flow state information is maintained at each switch, then congestion control precision improves, but memory requirements and switch complexity increase
Solution Approach 1:
The patent segments state information maintenance to only what is necessary for congestion control - per-flow queue states at each switch - rather than maintaining complete end-to-end flow state. This segmented approach provides sufficient precision for congestion detection while limiting memory requirements to local switch contexts.
Solution Approach 2:
Each switch independently maintains and updates its own per-flow state information based on local queue conditions and received packets. This self-service approach eliminates the need for centralized state management, reducing overall system complexity while maintaining precise local congestion detection capabilities.
3Productivity
If flow-specific input queues are provided at each switch, then traffic management efficiency improves, but buffer management complexity increases
Solution Approach 1:
The patent segments traffic into per-flow queues at each switch, enabling independent management and control of each flow's buffer. This segmentation improves traffic management efficiency by allowing selective congestion control on individual flows while the modular queue structure keeps buffer management complexity localized and manageable at each switch.
Data Source
Figure 1
Figure 2A~2B
Figure 3A~3C
AI summary
Data-driven intelligent networking systems and methods are provided. The system can accommodate dynamic traffic with fast, effective endpoint congestion detection and control. The system can maintain state information of individual packet flows, which can be set up or released dynamically based on injected data. Each flow can be provided with a flow-specific input queue upon arriving at a switch. Packets of a respective flow can be acknowledged after reaching the egress point of the network, and the acknowledgement packets can be sent back to the ingress point of the flow along the same data path. As a result, each switch can obtain state information of each flow and perform flow control on a per-flow basis.