A Dynamic Round Robin Grouping Scheduling Method with Good Fairness and High Efficiency

Abstract

The invention discloses a dynamic round-robin group scheduling method with good fairness and high efficiency. It includes the following steps: initialization; enqueue queuing process; dequeue scheduling process. During the out-of-queue scheduling process, it is sent according to the P value of the data flow in a polling manner. In the present invention, the P value of the data flow is calculated depending on the sending situation of each data flow in the previous round, so that the data flow that obtains little service in one round will be given more service opportunities in the next round , so as to ensure better fairness between data streams. The invention improves the efficiency of packet scheduling, fairly allocates bandwidth resources of output links, reduces congestion and packet loss rate, has low complexity, and at the same time, hardware implementation is also very simple.

Description

technical field [0001] The invention belongs to the technical field of network communication, and specifically relates to a dynamic polling group scheduling method with good fairness and high efficiency. Background technique [0002] With the development of the Internet, the services provided on the Internet are becoming more and more abundant. In addition to the traditional Best-effort service, many services represented by video conferencing, Internet telephony, and telemedicine services have been provided on the Internet. It is foreseeable that more businesses will be realized through the Internet in the future. Different services have different characteristics and have different quality requirements. For example, video conferencing has requirements for bandwidth and end-to-end delay, while traditional data services have no obvious QoS (Quality of Service) requirements. Therefore, even if the reliability and transmission rate of the transmission medium have been greatly i...

AI Technical Summary

Problems solved by technology

However, DRR has the following disadvantages: (1) it can only provide fairness on a long-term scale, and from a short-time scale, it will cause the output of the data stream to have a large burst; (2) the delay characteristic is not Ideally, some queues may not be served for a long time; (3) DRR assigns a quota value (Quantum) to each queue, and uses DC (Deficit Counter) to measure past unfairness. The packet will be sent only when the packet length is less than the sum of the quota value and DC, which requires that the packet length must be known before sending the packet, which increases the implementation overhead

Images