Dynamic Max-Min Fair Rate Regulation for Network Congestion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interconnection networks face challenges in optimizing data flow rates, leading to congestion and inefficient utilization of network resources due to the lack of effective rate regulation mechanisms that can manage peak traffic without relying on per-flow state or queues.
Innovation Solution
The Dynamic MAX-MIN FAIR RATE REGULATION system regulates data packet flow rates by calculating fair-share rates across network links, using feedback mechanisms to adjust source transmission rates, and distributing capacity fairly among flows, thereby preventing congestion and maximizing network utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If rate regulation mechanisms are implemented to manage peak traffic, then network congestion is prevented, but device complexity increases due to the need for fair-share rate calculation and feedback mechanisms
Solution Approach 1:
The system enables sources to self-regulate their transmission rates by autonomously calculating fair-share rates based on link capacity and flow count, and autonomously adjusting their injection rates according to feedback from rate regulation packets, eliminating the need for centralized control or complex switch-based regulation
Solution Approach 2:
The system implements feedback mechanisms where destinations send rate regulation packets containing current fair-share rate information back to sources, enabling sources to continuously adjust their transmission rates to match network conditions and prevent congestion
2Ease of operation
If per-flow state and queues are used for rate regulation, then precise flow control is achieved, but loss of time increases due to queue management overhead and state maintenance
Solution Approach 1:
The invention extracts and removes the queue management and per-flow state maintenance functions from the network switches and destinations, concentrating rate regulation logic at the source where flows are injected, thereby eliminating time-consuming queue operations and state updates in the network path
Solution Approach 2:
The system segments the rate regulation function from the data plane operations, separating flow control logic into distinct rate regulation packets that carry control information independently from data packets, allowing parallel processing without interference
3Productivity
If dynamic fair-share rate calculation is implemented, then network resource utilization is maximized, but loss of information increases due to the need for continuous rate reevaluation and signal updates
Solution Approach 1:
The system implements periodic rate reevaluation at defined intervals, where destinations calculate updated fair-share rates and send rate regulation packets to sources at regular periods, ensuring network resources are dynamically optimized while maintaining stable, predictable update cycles that prevent information loss
Solution Approach 2:
The system performs preliminary rate calculations and preparations before actual data transmission periods, allowing sources to be pre-configured with appropriate rate limits based on current network conditions, ensuring smooth transitions without disruption to ongoing data flows
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A processor-implemented method for regulating the flow rate of data packets in a network, including defining a global constant representing a regularly repeating time period common among flow sources in the network; transmitting current flow rate information from each of the flow sources, and for each flow, to the links traversed by each flow, exactly once during a current period; categorizing each of the flows passing through the links on the network into a category for the current period for each link by comparing the current flow rate information to a previously determined fair-share flow rate for the link; counting, in each link, the flows per category for the current period; determining a current fair-share flow rate for the current period in each link using the results of the categorizing and counting; and providing control instructions to each of the flow sources to regulate the rate of each flow.