VM Bandwidth Allocation for RoCE and TCP Congestion Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The challenge of efficiently managing bandwidth allocation between RoCE and TCP traffic on the same physical network in cloud computing platforms, as RoCE and TCP/IP protocols are incompatible, leading to inefficient use of computing resources and network communication overhead.

Innovation Solution

A method and apparatus for traffic congestion control that allocates network card bandwidth to virtual machines, dividing it into separate bandwidths for RoCE and TCP traffic, and reallocates based on actual usage, using an RDMA card with classifiers and meters to manage traffic types and prevent congestion.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If RoCE and TCP traffic share the same physical network, then network resource utilization is improved, but traffic congestion and protocol compatibility issues occur

Engineering Contradiction:
Improvenetwork resource utilizationVSAvoidtraffic transmission stability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent segments the shared physical network bandwidth into separate allocation pools for RoCE and TCP traffic. By dividing the network resources and implementing independent bandwidth management for each protocol type, the system allows both protocols to coexist on the same physical network while preventing mutual interference and congestion.

Inventive Principle:
Principle #1Segmentation

2Reliability

If bandwidth is statically allocated to RoCE and TCP traffic, then protocol compatibility is ensured, but bandwidth utilization efficiency decreases

Engineering Contradiction:
Improveprotocol compatibilityVSAvoidbandwidth utilization efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent implements dynamic bandwidth allocation that adjusts RoCE and TCP traffic bandwidth based on real-time network conditions and actual traffic demands. The system monitors network status continuously and reallocates bandwidth dynamically, ensuring both protocols maintain compatibility while maximizing overall bandwidth utilization efficiency.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent incorporates feedback mechanisms that monitor actual bandwidth usage and network congestion levels, then use this information to adjust bandwidth allocation for RoCE and TCP traffic. This closed-loop control ensures protocol compatibility is maintained while optimizing bandwidth utilization based on real-time conditions.

Inventive Principle:
Principle #23Feedback

3Speed

If more bandwidth is allocated to one protocol, then that protocol's performance is improved, but the other protocol experiences congestion

Engineering Contradiction:
Improveprotocol transmission speedVSAvoidtraffic congestion
Core Design Contradiction:
SpeedVSObject-affected harmful factors

Solution Approach 1:

The patent changes the bandwidth allocation parameters dynamically based on network conditions and traffic patterns. By adjusting the bandwidth parameters for RoCE and TCP traffic in real-time, the system prevents congestion in one protocol while maintaining or improving the transmission speed of the other, avoiding the zero-sum trade-off.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS12542743B2Method for traffic congestion control, computer-readable medium and electronic device
Publication Date: 2026.02.03 BEIJING YOUZHUJU NETWORK TECH CO LTD
  • US12542743B2 patent drawing
  • US12542743B2 patent drawing
  • US12542743B2 patent drawing

AI summary

The disclosure relates to a method, an apparatus for traffic congestion control, a computer-readable medium, and an electronic device. The method includes: allocating a network card bandwidth to a virtual machine on a same physical machine, where the network interface of the virtual machine is used for transmitting traffic of first and second transmission protocols; for any virtual machine on the physical machine, dividing a total bandwidth of the virtual machine into a first bandwidth, a second bandwidth, and a buffer bandwidth, where the first bandwidth is used to transmit traffic of the first transmission protocol, and the second bandwidth is used to transmit traffic of the second transmission protocol; and reallocating a bandwidth of the first and second transmission protocols based on the actual bandwidth used by the first and second transmission protocols.