Shared Buffer Service Differentiation via Modified Deficit Weighted Round Robin
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In network communications, managing different classes of service in a shared buffer is challenging due to congestion issues, where one class of traffic exceeding the drain rate can consume the entire buffer, invalidating service guarantees for other classes, especially when implementing a single shared buffer without per-class boundaries.
Innovation Solution
A modified deficit weighted round robin algorithm is used to control access to the buffer by issuing per-class Xon/Xoff signals and maintaining a traffic class profile vector, ensuring service guarantees by refreshing credit for classes with packets in the queue and preventing unicast lockout through careful credit management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single shared buffer is used without per-class boundaries, then buffer utilization efficiency is improved, but service guarantees for different traffic classes deteriorate
Solution Approach 1:
The patent segments the buffer management by implementing separate virtual queues for different traffic classes (voice, video, data) within the shared buffer. Each queue maintains independent tracking of packet counts and credit values, allowing the system to differentiate service guarantees while utilizing a single physical buffer resource. This segmentation enables the buffer to be shared physically while being managed logically as separate entities.
Solution Approach 2:
The patent changes the management parameters by introducing class-specific credit values and packet count tracking for each traffic type. The modified deficit weighted round robin algorithm uses these dynamic parameters to control buffer access, where each class accumulates credit at different rates and consumes buffer space according to its priority and traffic characteristics. This parameter-based control maintains service guarantees while enabling efficient shared buffer utilization.
2Reliability
If per-class queue management is implemented, then service differentiation is improved, but device complexity increases
Solution Approach 1:
The patent merges multiple per-class queue management functions into a single shared buffer structure. Instead of implementing separate physical queues for each traffic class, the system combines them into one buffer while maintaining logical separation through virtual queue pointers, credit tracking, and class-specific metadata. This merging reduces the complexity of managing multiple independent buffer structures while preserving service differentiation capabilities.
Solution Approach 2:
The shared buffer structure serves multiple functions simultaneously: it acts as a single storage resource for all traffic classes while also functioning as separate queues for voice, video, and data traffic. The buffer management system performs multiple tasks including credit accumulation, packet scheduling, and service guarantee enforcement within a unified structure, reducing overall system complexity compared to fully separate queue implementations.
3Productivity
If one traffic class exceeds drain rate and consumes entire buffer, then short-term throughput is improved, but service guarantees for other classes are invalidated
Solution Approach 1:
The patent implements feedback control through the modified deficit weighted round robin algorithm, which continuously monitors packet counts and credit values for each traffic class. When a class attempts to exceed its allocated buffer portion, the system detects this through credit depletion and adjusts scheduling decisions accordingly. This feedback mechanism prevents any single class from monopolizing the buffer while allowing legitimate throughput variations, maintaining service guarantees through dynamic adjustment based on real-time buffer state information.
Solution Approach 2:
The system performs preliminary action by pre-allocating credit values and establishing packet count limits for each traffic class before buffer consumption occurs. The deficit weighted round robin algorithm预先 calculates credit accumulation rates and scheduling priorities, preventing buffer monopolization before it can happen. This proactive control ensures that even if one class generates excessive traffic, the pre-established credit system limits its buffer consumption to maintain service guarantees for other classes.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An example method, system, and switching element are provided and may provide for an egress port to be configured to receive a plurality of data packets, each of the plurality of data packets being a class of a plurality of classes. A buffer may communicate with the at least one data port interface. A memory management unit may be configured to enable and disable transmission of the plurality of classes of the plurality of data packets based on a metering policy; and place the plurality of data packets in the buffer.