Network Controller Arbitration for HPC Traffic Class Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High Performance Computing (HPC) network fabrics struggle to accurately manage and prioritize different types of traffic based on Quality of Service (QoS) and latency requirements, leading to inefficiencies in bandwidth allocation and potential starvation of lower priority traffic.
Innovation Solution
A network controller is configured to arbitrate and preempt packets between multiple traffic classes using virtual lanes, adjusting priorities based on measured bandwidth consumption and latency sensitivity, ensuring higher-priority traffic does not deprive lower-priority traffic of bandwidth by implementing rules for bandwidth allocation and preemption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If higher priority traffic is given preferential treatment in bandwidth allocation, then latency and QoS for critical traffic are improved, but lower priority traffic may be starved of bandwidth
Solution Approach 1:
The patent implements dynamic priority adjustment where traffic class priorities are not fixed but can be modified in real-time based on current network conditions and bandwidth availability. The arbitration mechanism dynamically reassigns priorities to ensure lower priority traffic receives minimum guaranteed bandwidth while still allowing higher priority traffic to receive preferential treatment when bandwidth is available.
Solution Approach 2:
The system changes the parameter of traffic class priority dynamically rather than using static priority assignment. By monitoring bandwidth consumption and availability, the system adjusts priority parameters to balance QoS requirements with overall network utilization, preventing both starvation of lower priority traffic and excessive dominance by higher priority traffic.
2Device complexity
If existing network fabric switches approximate user configuration requests for priority and bandwidth share, then device complexity is reduced, but manufacturing precision and control accuracy deteriorate
Solution Approach 1:
The patent implements a feedback mechanism where the network switch continuously monitors actual bandwidth consumption by different traffic classes and compares it against allocated bandwidth shares. Based on this feedback, the arbitration logic adjusts priority assignments and bandwidth allocation in real-time, ensuring precise control over bandwidth distribution while maintaining a relatively simple switch architecture.
Solution Approach 2:
The arbitration mechanism performs self-adjustment based on monitored network conditions without requiring complex external control systems. The switch automatically detects bandwidth availability and traffic class requirements, then autonomously adjusts priority assignments to achieve precise bandwidth allocation while keeping the overall device complexity manageable.
Data Source
AI summary
This disclosure describes systems, devices, methods and computer readable media for enhanced network communication for use in higher performance applications including storage, high performance computing (HPC) and Ethernet-based fabric interconnects. In some embodiments, a network controller may include a transmitter circuit configured to transmit packets on a plurality of virtual lanes (VLs), the VLs associated with a defined VL priority and an allocated share of network bandwidth. The network controller may also include a bandwidth monitor module configured to measure bandwidth consumed by the packets and an arbiter module configured to adjust the VL priority based on a comparison of the measured bandwidth to the allocated share of network bandwidth. The transmitter circuit may be further configured to transmit the packets based on the adjusted VL priority.