Hierarchical weighted round-robin (WRR) scheduling device and method

Abstract

The invention provides a hierarchical weighted round-robin (WRR) scheduling device and method. The hierarchical WRR scheduling device comprises a first in first out (FIFO) cache queue module, a flow statistics module, a weight management module, and a scheduling control module, wherein the weight management module comprises a weight coefficient calculation sub-module, and a weight coefficient storage and updating sub-module. The hierarchical WRR scheduling method specifically comprises the following steps: 1) calculating a weight coefficient of each queue; 2) selecting a round-robin (RR) scheduling scheme; 3) carrying out RR scheduling based on the weight coefficients; and 4) carrying out hierarchical RR scheduling based on the weight coefficients. The device and method has the following advantages: 1) scheduling weights can be adjusted in a real-time manner according to business volumes of various priorities; 2) adaptation to different business distributions and business arrival rates can be achieved, a high service quality can be ensured, and the packet loss probability of businesses with high priorities can be reduced; and 3) the network construction cost can be reduced, and the network performance and the network service quality can be improved.

Description

technical field [0001] The invention relates to a layered weighted round robin scheduling device and method, belonging to the technical field of communication. Background technique [0002] With the increase of new functions and new scenarios of the Internet and the Internet of Things, various network applications are widely emerging. The switching network needs to be able to provide multi-priority business services. The traditional load-balancing Oueued) structure, it is generally believed that the queue can be of infinite length, which is inconsistent with the actual situation. When the input queue is of limited length, packet loss will occur. This problem is more obvious when scheduling multi-priority services. Traditional round-robin Inquiry scheduling is difficult to provide priority distinction, such as RR (Round-Robin) algorithm; if high-priority services are always prioritized, it will make it difficult for low-priority services to obtain services in time, such as ER...

AI Technical Summary

Problems solved by technology

[0005] RR method, which polls all queues in a fair manner, users use shared resources fairly, and cannot reflect the difference in service level between different priority queues;

[0006] In the WRR method, according to the priority of each queue, different weights are assigned to each queue, so that high-priority queues can get better services and avoid low-priority queues from occupying too much network bandwidth; however, each queue in WRR The weight value of the queue is fixedly allocated and cannot be adjusted dynamically with the arrival of business, so it may lead to low link utilization and poor flexibility

[0007] The ERR method always prioritizes high-priority queues, which will make it difficult for low-priority queue services to obtain timely services; the LQF method does not take into account the impact between different priority services, resulting in a large number of low-priority services. Priority services are better served, and the complexity is higher; the above existing queue scheduling methods all have the disadvantages of occupying more output link bandwidth and lower link utilization. The high requirements of the network increase the cost of network construction and severely limit the improvement of network service quality and network performance.

Images