Flow Control Aggregation in Scalable Switching Fabrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Packet-switched networks face challenges in managing flow control, particularly in scalable shared memory switching fabrics, where data loss occurs due to full switching fabrics and increased complexity in traffic control, exacerbated by clock rate differences between ingress and egress.
Innovation Solution
A flow control mechanism is implemented using a flow control aggregation unit that multiplexes flow control messages with other traffic, aggregating cells from queues and employing scheduling algorithms to manage bandwidth, along with write and read counters to maintain accurate queue fullness, and clock comparison to address clock rate disparities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the switching fabric size is increased to handle more traffic, then the system capacity is improved, but the complexity of traffic control increases
Solution Approach 1:
The patent segments the flow control process into distinct components: ingress modules generate flow control messages independently, the switching fabric processes cells independently, and egress modules aggregate flow control messages before sending feedback. This segmentation allows the system to scale capacity while keeping individual control units simple.
Solution Approach 2:
The patent merges multiple flow control messages at egress modules before sending feedback to the ingress. Instead of sending individual flow control messages for each cell, multiple messages are aggregated into a single feedback message, reducing control overhead and simplifying traffic control as system capacity increases.
2Measurement precision
If flow control messages are sent for each cell dequeued, then flow control accuracy is improved, but the overhead increases
Solution Approach 1:
Instead of sending flow control messages for every single cell dequeued, the system uses periodic aggregation where multiple flow control updates are combined into periodic feedback messages. This maintains flow control accuracy by tracking each cell while reducing overhead by batching the feedback transmissions.
Solution Approach 2:
The patent applies asymmetric treatment to flow control messages versus data cells. Individual cell dequeuing operations are tracked precisely, but the feedback mechanism uses aggregated, asymmetric messaging patterns where multiple cell events trigger a single consolidated flow control message, reducing overall overhead.
3Productivity
If the switching fabric operates at high speed, then productivity is improved, but clock rate differences between ingress and egress cause control difficulties
Solution Approach 1:
The flow control mechanism operates autonomously at each module without requiring centralized clock synchronization. Ingress modules independently generate flow control messages based on their local queue states, and egress modules independently aggregate and send feedback, allowing high-speed operation with different clock rates at different locations.
Solution Approach 2:
The system uses continuous feedback loops where egress modules send flow control feedback messages to ingress modules based on their queue states. This feedback mechanism allows each module to self-adjust its operation based on real-time conditions, maintaining coordination without requiring synchronized clocks across the entire system.
Data Source
AI summary
A traffic control system and method with flow control aggregation. The system includes a switching fabric and an ingress module. The switching fabric includes read counters that are associated with a plurality of queues. The read counters represent an aggregated number of cells dequeued from respective queues since a previous flow control message (FCM) was sent to the ingress module. The read counters are reset when a FCM is created. The ingress module includes write counters that are associated with the queues. The write counters are incremented each time a cell is sent to the respective queues. The write counters are decremented in accordance with the FCM when the FCM is received. Also, read counters for one or more queues are aggregated into a single FCM.