Receiver-Based ECN Feedback for Precision Congestion Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing congestion control algorithms in datacenter networks face challenges in achieving high throughput, ultra-low latency, and network stability due to limitations in handling congestion accurately and efficiently.
Innovation Solution
The implementation of High Precision Congestion Control (HPCC) using receiver-based feedback mechanisms, where receivers determine and communicate the optimal transmit rates to senders based on precise link load information, thereby reducing the overhead of feedback packets and improving fairness among flows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing congestion control algorithms are used, then network stability is maintained, but throughput and latency performance are insufficient
Solution Approach 1:
The patent implements receiver-based feedback mechanisms where receivers send congestion notifications to senders using ECN (Explicit Congestion Notification) bits in packet headers. This feedback loop enables senders to adjust their transmit rates dynamically based on real-time network congestion conditions, achieving both high throughput and network stability through precise congestion response.
Solution Approach 2:
The patent dynamically changes the ECN notification rate parameter based on measured network conditions. By adjusting the rate at which receivers send ECN notifications, the system can optimize throughput under different congestion scenarios while maintaining stability, resolving the contradiction between productivity and reliability.
2Measurement precision
If feedback packets are sent frequently to achieve precise congestion control, then congestion response accuracy improves, but feedback overhead increases
Solution Approach 1:
The patent merges congestion feedback with existing data packet transmissions by using ECN bits in the packet headers themselves. This eliminates the need for separate feedback packets, achieving precise congestion detection without increasing overall network traffic overhead, as the congestion information is carried along with the data flow.
Solution Approach 2:
The ECN bit acts as an intermediary mechanism that conveys congestion information without requiring additional feedback packets. The receiver sets the ECN bit in outgoing packets to signal congestion conditions, allowing precise congestion communication while avoiding the overhead of dedicated feedback traffic.
3Productivity
If transmit rates are increased to improve throughput, then network utilization improves, but congestion and packet loss increase
Solution Approach 1:
The patent implements preliminary congestion response by having receivers proactively set ECN bits in packets before actual packet loss occurs. This early warning mechanism allows senders to reduce transmit rates preemptively, preventing congestion buildup and packet loss while maintaining high network utilization through smooth rate adjustment.
Data Source
AI summary
Examples described herein relate to a network agent, when operational, to: receive a packet, determine transmit rate-related information for a sender network device based at least on operational and telemetry information accumulated in the received packet, and transmit the transmit rate-related information to the sender network device. In some examples, the network agent includes a network device coupled to a server, a server, or a network device. In some examples, the operational and telemetry information comprises: telemetry information generated by at least one network device in a path from the sender network device to the network agent.