Adaptive Output Scheduler for Converged Network Routers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fiber optic transmission systems struggle to efficiently manage diverse types of traffic with varying service characteristics and priorities, leading to suboptimal bandwidth utilization and quality of service (QoS) issues, particularly in converged networks where high-priority real-time traffic and bursty background traffic coexist.
Innovation Solution
A programmable converged network router with adaptive output scheduling and bandwidth management, capable of executing multiple QoS algorithms, ensures differentiated services by allocating specific bandwidth and scheduling priorities, maximizing link utilization and preserving traffic characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If single stage crossbar switch fabric with centralized controller is used, then routing control is simplified, but performance bottleneck occurs and throughput is limited to 58.6% due to HOL blocking
Solution Approach 1:
The patent segments the centralized scheduling function into distributed scheduling units at each output port. Each output port independently schedules its own queue, eliminating the centralized controller bottleneck. This segmentation allows parallel processing of scheduling decisions and removes the single point of failure that limited throughput to 58.6%.
Solution Approach 2:
The patent transitions from input queuing to output queuing architecture. By placing queues at the output ports rather than input ports, the system eliminates head-of-line blocking at the input stage. This dimensional change in queue placement allows packets to be scheduled based on output port availability rather than input port constraints, achieving full throughput.
2Productivity
If virtual output queuing is implemented to eliminate HOL blocking, then throughput improves, but device complexity increases due to multiple virtual queues at each input port
Solution Approach 1:
The patent extracts the queuing function from the input ports and relocates it to the output ports. Instead of maintaining multiple virtual queues at each input port, the system uses a single physical queue at each output port. This extraction simplifies the input side while maintaining the throughput benefits of output queuing.
Solution Approach 2:
Each output port independently manages its own queue without requiring complex coordination with other ports. The distributed scheduling architecture allows each output port to autonomously make scheduling decisions based on local queue state and traffic demands, eliminating the need for complex global queue management mechanisms.
3Productivity
If CIOQ switch architecture is used, then bandwidth utilization improves, but unpredictable latency and jitter occur due to lack of QOS guarantees
Solution Approach 1:
The patent implements differentiated service qualities at each output port through independent scheduling. Each output port can apply different scheduling algorithms and QOS policies tailored to its specific traffic requirements. This local quality control allows simultaneous support for real-time traffic with strict latency requirements and best-effort traffic with flexible timing, achieving both high utilization and predictable latency for priority traffic.
Solution Approach 2:
The patent employs dynamic scheduling algorithms that adapt to changing traffic conditions in real-time. The scheduling decisions are made based on current queue states, traffic priorities, and QOS requirements rather than static configurations. This dynamic approach allows the system to maintain predictable latency for high-priority traffic while maximizing overall bandwidth utilization under varying load conditions.
4Reliability
If strict priority scheduling is applied, then real-time traffic QOS is guaranteed, but bandwidth waste occurs when high-priority queues have no packets to transmit
Solution Approach 1:
The patent implements a periodic round-robin scheduling mechanism that cycles through different priority queues. When high-priority queues are empty, the scheduler naturally transitions to lower-priority queues in a periodic fashion, ensuring that bandwidth is continuously utilized without waste. This periodic action maintains QOS guarantees by always checking high-priority queues first while efficiently utilizing available bandwidth from lower-priority traffic during idle periods.
Solution Approach 2:
The patent applies weighted fair queuing that allocates bandwidth proportionally based on traffic priorities rather than giving absolute priority to high-priority traffic. This partial priority approach ensures that high-priority traffic receives sufficient bandwidth to meet QOS requirements while allowing lower-priority traffic to utilize remaining bandwidth, preventing waste when high-priority queues are empty.
Data Source
AI summary
A configurable adaptive variable length data packet transmission output scheduler for enabling substantially simultaneous transmission on a common transmission link, as of fiber optics, of differentiated services for various different traffic types, executing different QOS algorithms while co-existing in a converged network environment, with simultaneous preserving of the different service characteristics for real-time or high-priority traffic and providing differentiated bandwidth allocation while achieving maximal link utilization—all through a fine and balanced control as to which type of traffic is transmitted on the link for a given duration, and how much of that traffic is transmitted on the link.