分组调度算法是路由交换设备性能的重要保证,对基于轮询的分组调度进行了研究,提出了一种新的调度算法称为逐次最小权值轮询调度算法(successive minimal-weight round robin,SMRR),在每个轮次中为每个活动数据流提供与本轮次中的最小权值相当的服务机会.根据Latency-Rate(LR)Servers理论,证明了SMRR算法和WRR算法的时延上界,并对SMRR算法的公平性和实现复杂性进行了讨论,理论推导和性能分析表明SMRR算法具有比WRR算法更好的时延特性和公平性,同时具有O(1)的时间复杂度,具有良好的可扩展性.
Packet scheduling algorithms are playing a significant role in guaranteeing the performance of routing and switching devices.Researching on round-robin-based packet scheduling,this paper presents a new scheduling algorithm called Successive Minimal-weight Round Robin(SMRR),which is fair,efficient and has a low latency bound.The main idea of SMRR is:In each round,SMRR offers the consistent service opportunity,which is corresponding to the minimal weight of the current round,for all active flows.According to the theory of Latency-Rate(LR) servers introduced by Stiliadis and Varma,this paper proves the upper bound on the latency of SMRR and WRR,and discusses the fairness and implementation complexity of SMRR.Theoretical derivation and performance analysis shows that SMRR algorithm has better latency characteristics and fairness than WRR,and simultaneously possesses the time complexity of O(1) with the number of flows and good scalability.