论文部分内容阅读
在差别服务网络中,实时业务的延时保证是一个难题,这需要在调度聚集业务的时候能够保存聚集类中每一个单独流的业务特性。针对此问题,该文提出了一种用于差别服务网络的调度与整形相结合的机制——TD2FQ(Token-drivenDelay-sensitiveDynamicFairQueueing)。该机制包括一种新的调度机制——延时相关动态公平队列,以及一种新的“有色令牌”应用模块。TD2FQ可用于进行聚集业务的成形调度,除了能够获得确保的延时上界,还可以维持业务类中不同流间的公平性以及业务流的成形条件,并且保证了输出链路的带宽利用率。该文给出了该算法的理论分析和仿真结果,论证了该算法的优越性。
In differentiated services networks, real-time service delay guarantee is a challenge, which needs to be able to save the business characteristics of each individual flow in the aggregation class when scheduling the aggregated services. In order to solve this problem, this paper proposes Token-drivenDelay-sensitiveDynamicFairQueueing (TDDF), which is a combination of scheduling and shaping for differentiated services networks. The mechanism includes a new scheduling mechanism - delay related dynamic fair queuing, and a new “tokens” application module. TD2FQ can be used to perform shaping scheduling of aggregated services. In addition to ensuring the upper bound of delay, TD2FQ can also maintain the fairness of different streams in the service class and shaping conditions of the service flow, and ensure the bandwidth utilization of the output link. The paper gives the theoretical analysis and simulation results of the algorithm, and demonstrates the superiority of the algorithm.