Tree Topology Congestion Control via Service Queue Bandwidth Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current congestion control methods in chain topology networks are ineffective in tree topology networks, as they reduce the available bandwidth of uncongested links, affecting the overall network throughput.
Innovation Solution
A congestion control method for tree topology networks that involves a root node network element receiving link notification messages, determining congested links, and adjusting the output bandwidth of service queues to alleviate congestion without affecting uncongested links, using a correspondence table to identify leaf node network elements and service queues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the root node network element reduces output bandwidth of a port to control congestion on one link, then the congestion on the congested link is alleviated, but the available bandwidth of other uncongested links connected to the same intermediate node is reduced, affecting overall network throughput
Solution Approach 1:
The patent segments the bandwidth control by introducing the concept of service queues corresponding to different leaf node network elements. Instead of controlling all traffic through an intermediate node uniformly, the root node network element identifies and controls only the service queue corresponding to the congested link's leaf node. This segmentation allows uncongested links to maintain their bandwidth while the congested link receives targeted bandwidth reduction.
Solution Approach 2:
The patent applies local quality by making the bandwidth control specific to the congested link rather than applying uniform control to all links. The root node network element determines which service queue corresponds to the congested link and adjusts only that specific queue's output bandwidth. This localized control ensures that bandwidth is reduced only where congestion occurs, preserving throughput on uncongested links.
2Reliability
If the root node network element uses minimum value of available bandwidth across all links to control congestion, then congestion on any link is controlled, but bandwidth of uncongested links is unnecessarily reduced
Solution Approach 1:
The patent makes bandwidth control link-specific by identifying the service queue corresponding to the congested link. Instead of using the minimum available bandwidth across all links (which would unnecessarily reduce bandwidth on uncongested links), the root node network element adjusts only the output bandwidth of the specific service queue associated with the congested link. This localized approach prevents unnecessary bandwidth loss on uncongested links while still controlling congestion where it occurs.
3Adaptability or versatility
If the congestion control method is designed for chain topology networks, then it works effectively for linear link structures, but it cannot resolve congestion problems in tree topology networks where intermediate nodes connect to multiple other intermediate nodes
Solution Approach 1:
The patent makes the congestion control method universal by introducing the correspondence relationship between service queues and leaf node network elements. This approach works for both chain topology networks (where each intermediate node connects to at most one other intermediate node) and tree topology networks (where intermediate nodes can connect to multiple other intermediate nodes). The root node network element uses the correspondence table to identify the correct service queue regardless of the network topology, making the method applicable to multiple network structures.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention discloses a congestion control method and an apparatus, relates to the field of communications technologies, and is applied to a tree topology network, so as to alleviate congestion of a congested link without affecting bandwidth of another uncongested link, thereby optimizing a throughput of the entire network. The method includes: receiving, by a root node network element, a first link notification message that is sent by a first intermediate node network element and that carries at least link status information of a first link that is directly connected to a first port of the first intermediate node network element; determining, based on the link status information of the first link, that the first link is congested; determining an identifier of a first leaf node network element based on a correspondence between an identifier of the first port and the identifier of the first leaf node network element; determining a service queue of the first leaf node network element based on a correspondence between the identifier of the first leaf node network element and an identifier of the service queue of the first leaf node network element; and lowering output bandwidth of the service queue of the first leaf node network element.