Hardware Rate Control Engine for Bursty Traffic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hardware-based rate controllers struggle with bursty traffic, leading to packet drops and increased bandwidth usage due to a 'use-it or lose-it' credit scheme, which is inefficient for TCP traffic and requires dedicated buffer memory, and are inflexible with changing traffic patterns.
Innovation Solution
A hardware-based rate control engine that accumulates credits over multiple time-slices with a maximum credit limit and adapts settings based on traffic characteristics, allowing flexible rate control without packet buffering, enabling efficient handling of bursty traffic and varying traffic patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a use-it or lose-it credit scheme is used to control packet traffic, then packet traffic can be limited to the rate established by the refresh rate, but packets from bursty traffic must be dropped when they exceed the allocated credits in the current time interval
Solution Approach 1:
The patent applies dynamics by making the credit bucket system adaptable to different traffic patterns. The rate controller dynamically adjusts its behavior based on whether traffic is classified as elastic or inelastic, allowing credits to accumulate during low-traffic periods for elastic traffic while maintaining strict rate limiting for inelastic traffic. This dynamic adaptation resolves the contradiction by preventing packet drops for bursty elastic traffic while maintaining rate control effectiveness.
Solution Approach 2:
The patent changes parameters by introducing traffic pattern classification (elastic vs. inelastic) that modifies how credits are managed. For elastic traffic, the system allows credit accumulation beyond the current time interval's allocation, effectively changing the rate control parameter from strict per-interval limiting to cumulative rate shaping. This parameter change eliminates packet drops for bursty traffic while maintaining overall rate control.
2Loss of substance
If packets are buffered to allow rate shaping instead of rate limiting, then packet drops can be reduced, but buffer memory adds cost and adds delay to packets
Solution Approach 1:
The patent extracts the buffering function from physical memory and implements it virtually through the credit accumulation mechanism. Instead of requiring actual buffer memory to hold packets during rate shaping, the system uses credit buckets to represent available bandwidth capacity. This extraction eliminates the need for dedicated buffer memory while achieving the same rate shaping effect, resolving the contradiction between reducing packet drops and avoiding buffer memory costs.
Solution Approach 2:
The patent substitutes the mechanical buffering system with a virtual credit-based control system. Rather than physically storing packets in memory buffers, the system uses software-based credit accumulation and tracking to achieve rate shaping. This substitution eliminates the hardware complexity of buffer memory while maintaining the ability to smooth bursty traffic without drops.
3Ease of operation
If rate control parameters are set once and left alone, then the system is simple to operate, but it does not work well when the traffic pattern is unpredictable
Solution Approach 1:
The patent implements feedback by continuously monitoring traffic patterns and automatically classifying traffic as elastic or inelastic based on observed behavior. The system uses this feedback to dynamically adjust rate control behavior without requiring manual parameter reconfiguration. This feedback mechanism resolves the contradiction by maintaining ease of operation (no manual tuning needed) while achieving high adaptability to changing traffic patterns through automatic classification and response.
Solution Approach 2:
The patent applies self-service by enabling the rate controller to automatically adapt to different traffic patterns without external intervention. The system autonomously classifies traffic flows, selects appropriate rate control strategies, and adjusts parameters as needed. This self-service capability resolves the contradiction by eliminating the need for manual parameter adjustment while providing sophisticated adaptability to various traffic patterns.
Data Source
AI summary
A hardware-based rate control engine is designed to allow credits to be accumulated over multiple time-slices up to a maximum credit limit. The rate control engine is also allowed to dispatch packets up to a maximum drain rate during each time-slice as long as sufficient credits are available. Allowing credits to accumulate over multiple time slices allows unused bandwidth to be saved during times of little or no traffic and used when the subsequent traffic bursts arrive. Additionally, limiting the maximum drain rate puts a cap on how fast the saved bandwidth can be consumed by subsequent traffic bursts. The rate control settings can be adapted in response to a characteristic, or characteristics, of the traffic flow. For example, one group of settings of a traffic flow can be used for bursty traffic, while a different group of settings can be used for smooth traffic.