Network System Dynamic Data Flow Control via Software Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional network systems supporting differentiated services require a large number of physical output queues to manage different service classes, leading to inefficiencies in packet management and increased costs.
Innovation Solution
A network system with a forwarding unit and a processor that dynamically controls data flow by classifying packets and adjusting parameters using software, reducing the need for multiple physical output queues while maintaining quality of service for different service classes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical output queues are established as many as the number of service classes, then different service classes can be managed separately, but the device complexity and hardware costs increase significantly
Solution Approach 1:
The patent merges multiple service class queues into a single shared physical output queue. Instead of having separate physical queues for each service class (EF, AF1, AF2, AF3, BE), all service classes share one physical queue. The differentiation is achieved through software-based priority scheduling that dynamically allocates bandwidth and manages packet transmission order, thus reducing hardware complexity while maintaining service class management capability.
Solution Approach 2:
The patent replaces the mechanical/physical queue structure with a software-based scheduling system. The forwarding unit uses a priority scheduler that dynamically assigns transmission priorities to different service classes through software control. This substitution eliminates the need for multiple physical queues while maintaining the ability to differentiate and manage service classes, directly addressing the contradiction between reliability of service class management and device complexity.
2Ease of operation
If multiple physical output queues are used for different service classes, then packet management can be performed separately, but the hardware costs and resource utilization efficiency decrease
Solution Approach 1:
The single physical output queue is designed to serve multiple service classes simultaneously through software-based priority scheduling. The queue structure is universal and can handle packets from different service classes (EF, AF1, AF2, AF3, BE) without requiring dedicated physical queues for each class. This multi-functionality reduces hardware resource consumption while maintaining separate packet management capabilities through the scheduler.
Solution Approach 2:
The patent implements dynamic priority scheduling where the transmission priority of different service classes can be adjusted in real-time based on network conditions and service requirements. The priority scheduler dynamically allocates bandwidth and manages packet transmission order, allowing flexible packet management without the rigidity of fixed physical queue assignments. This dynamic approach improves hardware resource utilization while maintaining ease of packet management.
3Adaptability or versatility
If a conventional Diffserv domain is used, then service classes can be classified, but the core router cannot efficiently manage packets without additional forwarding information for each route
Solution Approach 1:
The patent extracts the packet management functionality from the core router and relocates it to the edge router. The edge router performs service class classification and applies priority scheduling, while the core router simply forwards packets based on standard routing information. This extraction eliminates the need for the core router to store and process additional forwarding information for each service class route, maintaining service class adaptability while reducing core router complexity.
Solution Approach 2:
The edge router acts as an intermediary that handles service class classification and priority scheduling before packets enter the core Diffserv domain. By placing the intelligent scheduling function at the edge, the core router can operate with simple forwarding information without needing to understand or manage multiple service classes. This intermediary approach maintains service class versatility while avoiding the complexity of extended forwarding information requirements in the core.
Data Source
AI summary
A network system capable of dynamically controlling a data flow is disclosed. The network system includes: a forwarding unit, for receiving a packet and for providing a QoS (quality of service) function; and a processor, for accessing data from the forwarding unit through an interface and setting at least one parameter of the forwarding unit. The interface is coupled between the forwarding unit and the processor, and the processor executes a software program to set the at least one parameter of the forwarding unit.In this way, the forwarding unit can classify received packets into different QoS levels according to their transmission properties, and can provide different forwarding methods and QoS services to reduce the amount of the output queues physically.