Method and device for queue management

Abstract

The invention provides a method and a device for queue management. The method for queue management comprises the steps of: determining a first logical queue indicated by a first logic queue head pointer and a second logic queue indicated by a second logic queue head pointer according to a queue validity identifier corresponding to an entity queue when a message enters the entity queue, wherein the first logic queue and the second logic queue comprise the same tail pointers; reading a scheduling parameter information corresponding to the first logic queue head pointer when the first logic queue dequeues, and reading a message descriptor according to a final scheduling result obtained through the scheduling parameter information when the second logic queue dequeues. Thus, the special function requirement that a push-type multi-level scheduler can schedule and select the queue after obtaining the scheduling parameter information firstly before scheduling is met, and the message descriptor only can be read from a message descriptor table when the scheduler outputs the final scheduling result. Therefore, cache resources are greatly saved.

Description

technical field [0001] The embodiments of the present invention relate to communication service quality of service, and more specifically, to a queue management method and device. Background technique [0002] QM (Queue Manager, Queue Manager), as a core functional processing unit for realizing QoS (Quality of Service, Quality of Service) characteristics of communication services, needs to cooperate with a scheduler (scheduler) that applies various scheduling algorithms. There are two basic implementations of the multi-level scheduler: pull (from root to leaf) and push (from leaf to root). [0003] Some application scenarios may have higher requirements on the fairness index of the scheduling algorithm, for example, it is necessary to select a real WFQ (Weighted Fair Queuing, Weighted Fair Queuing) algorithm based on virtual time scale comparison. When implementing a scheduler that is closer to the original definition of WFQ, the scheduler needs to first obtain the packet l...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a queue management method and a queue management device, which can solve the problem of the traditional one-head-one-tail single-linked list queue manager needing to repeatedly store part of the table item content in the on-chip memory when it cooperates with the push-type multi-level scheduler. The problem with a large number of on-chip cache resources

Images